Merge branch 'ss/get-time-cleanup'
[git/git.git] / diff.c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "config.h"
6 #include "tempfile.h"
7 #include "quote.h"
8 #include "diff.h"
9 #include "diffcore.h"
10 #include "delta.h"
11 #include "xdiff-interface.h"
12 #include "color.h"
13 #include "attr.h"
14 #include "run-command.h"
15 #include "utf8.h"
16 #include "object-store.h"
17 #include "userdiff.h"
18 #include "submodule-config.h"
19 #include "submodule.h"
20 #include "hashmap.h"
21 #include "ll-merge.h"
22 #include "string-list.h"
23 #include "argv-array.h"
24 #include "graph.h"
25 #include "packfile.h"
26 #include "parse-options.h"
27 #include "help.h"
28 #include "promisor-remote.h"
29
30 #ifdef NO_FAST_WORKING_DIRECTORY
31 #define FAST_WORKING_DIRECTORY 0
32 #else
33 #define FAST_WORKING_DIRECTORY 1
34 #endif
35
36 static int diff_detect_rename_default;
37 static int diff_indent_heuristic = 1;
38 static int diff_rename_limit_default = 400;
39 static int diff_suppress_blank_empty;
40 static int diff_use_color_default = -1;
41 static int diff_color_moved_default;
42 static int diff_color_moved_ws_default;
43 static int diff_context_default = 3;
44 static int diff_interhunk_context_default;
45 static const char *diff_word_regex_cfg;
46 static const char *external_diff_cmd_cfg;
47 static const char *diff_order_file_cfg;
48 int diff_auto_refresh_index = 1;
49 static int diff_mnemonic_prefix;
50 static int diff_no_prefix;
51 static int diff_stat_graph_width;
52 static int diff_dirstat_permille_default = 30;
53 static struct diff_options default_diff_options;
54 static long diff_algorithm;
55 static unsigned ws_error_highlight_default = WSEH_NEW;
56
57 static char diff_colors[][COLOR_MAXLEN] = {
58 GIT_COLOR_RESET,
59 GIT_COLOR_NORMAL, /* CONTEXT */
60 GIT_COLOR_BOLD, /* METAINFO */
61 GIT_COLOR_CYAN, /* FRAGINFO */
62 GIT_COLOR_RED, /* OLD */
63 GIT_COLOR_GREEN, /* NEW */
64 GIT_COLOR_YELLOW, /* COMMIT */
65 GIT_COLOR_BG_RED, /* WHITESPACE */
66 GIT_COLOR_NORMAL, /* FUNCINFO */
67 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
68 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
69 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
70 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
71 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
72 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
73 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
74 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
75 GIT_COLOR_FAINT, /* CONTEXT_DIM */
76 GIT_COLOR_FAINT_RED, /* OLD_DIM */
77 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
78 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
79 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
80 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
81 };
82
83 static const char *color_diff_slots[] = {
84 [DIFF_CONTEXT] = "context",
85 [DIFF_METAINFO] = "meta",
86 [DIFF_FRAGINFO] = "frag",
87 [DIFF_FILE_OLD] = "old",
88 [DIFF_FILE_NEW] = "new",
89 [DIFF_COMMIT] = "commit",
90 [DIFF_WHITESPACE] = "whitespace",
91 [DIFF_FUNCINFO] = "func",
92 [DIFF_FILE_OLD_MOVED] = "oldMoved",
93 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
94 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
95 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
96 [DIFF_FILE_NEW_MOVED] = "newMoved",
97 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
98 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
99 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
100 [DIFF_CONTEXT_DIM] = "contextDimmed",
101 [DIFF_FILE_OLD_DIM] = "oldDimmed",
102 [DIFF_FILE_NEW_DIM] = "newDimmed",
103 [DIFF_CONTEXT_BOLD] = "contextBold",
104 [DIFF_FILE_OLD_BOLD] = "oldBold",
105 [DIFF_FILE_NEW_BOLD] = "newBold",
106 };
107
108 define_list_config_array_extra(color_diff_slots, {"plain"});
109
110 static int parse_diff_color_slot(const char *var)
111 {
112 if (!strcasecmp(var, "plain"))
113 return DIFF_CONTEXT;
114 return LOOKUP_CONFIG(color_diff_slots, var);
115 }
116
117 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
118 struct strbuf *errmsg)
119 {
120 char *params_copy = xstrdup(params_string);
121 struct string_list params = STRING_LIST_INIT_NODUP;
122 int ret = 0;
123 int i;
124
125 if (*params_copy)
126 string_list_split_in_place(&params, params_copy, ',', -1);
127 for (i = 0; i < params.nr; i++) {
128 const char *p = params.items[i].string;
129 if (!strcmp(p, "changes")) {
130 options->flags.dirstat_by_line = 0;
131 options->flags.dirstat_by_file = 0;
132 } else if (!strcmp(p, "lines")) {
133 options->flags.dirstat_by_line = 1;
134 options->flags.dirstat_by_file = 0;
135 } else if (!strcmp(p, "files")) {
136 options->flags.dirstat_by_line = 0;
137 options->flags.dirstat_by_file = 1;
138 } else if (!strcmp(p, "noncumulative")) {
139 options->flags.dirstat_cumulative = 0;
140 } else if (!strcmp(p, "cumulative")) {
141 options->flags.dirstat_cumulative = 1;
142 } else if (isdigit(*p)) {
143 char *end;
144 int permille = strtoul(p, &end, 10) * 10;
145 if (*end == '.' && isdigit(*++end)) {
146 /* only use first digit */
147 permille += *end - '0';
148 /* .. and ignore any further digits */
149 while (isdigit(*++end))
150 ; /* nothing */
151 }
152 if (!*end)
153 options->dirstat_permille = permille;
154 else {
155 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
156 p);
157 ret++;
158 }
159 } else {
160 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
161 ret++;
162 }
163
164 }
165 string_list_clear(&params, 0);
166 free(params_copy);
167 return ret;
168 }
169
170 static int parse_submodule_params(struct diff_options *options, const char *value)
171 {
172 if (!strcmp(value, "log"))
173 options->submodule_format = DIFF_SUBMODULE_LOG;
174 else if (!strcmp(value, "short"))
175 options->submodule_format = DIFF_SUBMODULE_SHORT;
176 else if (!strcmp(value, "diff"))
177 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
178 /*
179 * Please update $__git_diff_submodule_formats in
180 * git-completion.bash when you add new formats.
181 */
182 else
183 return -1;
184 return 0;
185 }
186
187 int git_config_rename(const char *var, const char *value)
188 {
189 if (!value)
190 return DIFF_DETECT_RENAME;
191 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
192 return DIFF_DETECT_COPY;
193 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
194 }
195
196 long parse_algorithm_value(const char *value)
197 {
198 if (!value)
199 return -1;
200 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
201 return 0;
202 else if (!strcasecmp(value, "minimal"))
203 return XDF_NEED_MINIMAL;
204 else if (!strcasecmp(value, "patience"))
205 return XDF_PATIENCE_DIFF;
206 else if (!strcasecmp(value, "histogram"))
207 return XDF_HISTOGRAM_DIFF;
208 /*
209 * Please update $__git_diff_algorithms in git-completion.bash
210 * when you add new algorithms.
211 */
212 return -1;
213 }
214
215 static int parse_one_token(const char **arg, const char *token)
216 {
217 const char *rest;
218 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
219 *arg = rest;
220 return 1;
221 }
222 return 0;
223 }
224
225 static int parse_ws_error_highlight(const char *arg)
226 {
227 const char *orig_arg = arg;
228 unsigned val = 0;
229
230 while (*arg) {
231 if (parse_one_token(&arg, "none"))
232 val = 0;
233 else if (parse_one_token(&arg, "default"))
234 val = WSEH_NEW;
235 else if (parse_one_token(&arg, "all"))
236 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
237 else if (parse_one_token(&arg, "new"))
238 val |= WSEH_NEW;
239 else if (parse_one_token(&arg, "old"))
240 val |= WSEH_OLD;
241 else if (parse_one_token(&arg, "context"))
242 val |= WSEH_CONTEXT;
243 else {
244 return -1 - (int)(arg - orig_arg);
245 }
246 if (*arg)
247 arg++;
248 }
249 return val;
250 }
251
252 /*
253 * These are to give UI layer defaults.
254 * The core-level commands such as git-diff-files should
255 * never be affected by the setting of diff.renames
256 * the user happens to have in the configuration file.
257 */
258 void init_diff_ui_defaults(void)
259 {
260 diff_detect_rename_default = DIFF_DETECT_RENAME;
261 }
262
263 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
264 {
265 if (!strcmp(var, "diff.indentheuristic"))
266 diff_indent_heuristic = git_config_bool(var, value);
267 return 0;
268 }
269
270 static int parse_color_moved(const char *arg)
271 {
272 switch (git_parse_maybe_bool(arg)) {
273 case 0:
274 return COLOR_MOVED_NO;
275 case 1:
276 return COLOR_MOVED_DEFAULT;
277 default:
278 break;
279 }
280
281 if (!strcmp(arg, "no"))
282 return COLOR_MOVED_NO;
283 else if (!strcmp(arg, "plain"))
284 return COLOR_MOVED_PLAIN;
285 else if (!strcmp(arg, "blocks"))
286 return COLOR_MOVED_BLOCKS;
287 else if (!strcmp(arg, "zebra"))
288 return COLOR_MOVED_ZEBRA;
289 else if (!strcmp(arg, "default"))
290 return COLOR_MOVED_DEFAULT;
291 else if (!strcmp(arg, "dimmed-zebra"))
292 return COLOR_MOVED_ZEBRA_DIM;
293 else if (!strcmp(arg, "dimmed_zebra"))
294 return COLOR_MOVED_ZEBRA_DIM;
295 else
296 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
297 }
298
299 static unsigned parse_color_moved_ws(const char *arg)
300 {
301 int ret = 0;
302 struct string_list l = STRING_LIST_INIT_DUP;
303 struct string_list_item *i;
304
305 string_list_split(&l, arg, ',', -1);
306
307 for_each_string_list_item(i, &l) {
308 struct strbuf sb = STRBUF_INIT;
309 strbuf_addstr(&sb, i->string);
310 strbuf_trim(&sb);
311
312 if (!strcmp(sb.buf, "no"))
313 ret = 0;
314 else if (!strcmp(sb.buf, "ignore-space-change"))
315 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
316 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
317 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
318 else if (!strcmp(sb.buf, "ignore-all-space"))
319 ret |= XDF_IGNORE_WHITESPACE;
320 else if (!strcmp(sb.buf, "allow-indentation-change"))
321 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
322 else {
323 ret |= COLOR_MOVED_WS_ERROR;
324 error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
325 }
326
327 strbuf_release(&sb);
328 }
329
330 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
331 (ret & XDF_WHITESPACE_FLAGS)) {
332 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
333 ret |= COLOR_MOVED_WS_ERROR;
334 }
335
336 string_list_clear(&l, 0);
337
338 return ret;
339 }
340
341 int git_diff_ui_config(const char *var, const char *value, void *cb)
342 {
343 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
344 diff_use_color_default = git_config_colorbool(var, value);
345 return 0;
346 }
347 if (!strcmp(var, "diff.colormoved")) {
348 int cm = parse_color_moved(value);
349 if (cm < 0)
350 return -1;
351 diff_color_moved_default = cm;
352 return 0;
353 }
354 if (!strcmp(var, "diff.colormovedws")) {
355 unsigned cm = parse_color_moved_ws(value);
356 if (cm & COLOR_MOVED_WS_ERROR)
357 return -1;
358 diff_color_moved_ws_default = cm;
359 return 0;
360 }
361 if (!strcmp(var, "diff.context")) {
362 diff_context_default = git_config_int(var, value);
363 if (diff_context_default < 0)
364 return -1;
365 return 0;
366 }
367 if (!strcmp(var, "diff.interhunkcontext")) {
368 diff_interhunk_context_default = git_config_int(var, value);
369 if (diff_interhunk_context_default < 0)
370 return -1;
371 return 0;
372 }
373 if (!strcmp(var, "diff.renames")) {
374 diff_detect_rename_default = git_config_rename(var, value);
375 return 0;
376 }
377 if (!strcmp(var, "diff.autorefreshindex")) {
378 diff_auto_refresh_index = git_config_bool(var, value);
379 return 0;
380 }
381 if (!strcmp(var, "diff.mnemonicprefix")) {
382 diff_mnemonic_prefix = git_config_bool(var, value);
383 return 0;
384 }
385 if (!strcmp(var, "diff.noprefix")) {
386 diff_no_prefix = git_config_bool(var, value);
387 return 0;
388 }
389 if (!strcmp(var, "diff.statgraphwidth")) {
390 diff_stat_graph_width = git_config_int(var, value);
391 return 0;
392 }
393 if (!strcmp(var, "diff.external"))
394 return git_config_string(&external_diff_cmd_cfg, var, value);
395 if (!strcmp(var, "diff.wordregex"))
396 return git_config_string(&diff_word_regex_cfg, var, value);
397 if (!strcmp(var, "diff.orderfile"))
398 return git_config_pathname(&diff_order_file_cfg, var, value);
399
400 if (!strcmp(var, "diff.ignoresubmodules"))
401 handle_ignore_submodules_arg(&default_diff_options, value);
402
403 if (!strcmp(var, "diff.submodule")) {
404 if (parse_submodule_params(&default_diff_options, value))
405 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
406 value);
407 return 0;
408 }
409
410 if (!strcmp(var, "diff.algorithm")) {
411 diff_algorithm = parse_algorithm_value(value);
412 if (diff_algorithm < 0)
413 return -1;
414 return 0;
415 }
416
417 if (!strcmp(var, "diff.wserrorhighlight")) {
418 int val = parse_ws_error_highlight(value);
419 if (val < 0)
420 return -1;
421 ws_error_highlight_default = val;
422 return 0;
423 }
424
425 if (git_color_config(var, value, cb) < 0)
426 return -1;
427
428 return git_diff_basic_config(var, value, cb);
429 }
430
431 int git_diff_basic_config(const char *var, const char *value, void *cb)
432 {
433 const char *name;
434
435 if (!strcmp(var, "diff.renamelimit")) {
436 diff_rename_limit_default = git_config_int(var, value);
437 return 0;
438 }
439
440 if (userdiff_config(var, value) < 0)
441 return -1;
442
443 if (skip_prefix(var, "diff.color.", &name) ||
444 skip_prefix(var, "color.diff.", &name)) {
445 int slot = parse_diff_color_slot(name);
446 if (slot < 0)
447 return 0;
448 if (!value)
449 return config_error_nonbool(var);
450 return color_parse(value, diff_colors[slot]);
451 }
452
453 /* like GNU diff's --suppress-blank-empty option */
454 if (!strcmp(var, "diff.suppressblankempty") ||
455 /* for backwards compatibility */
456 !strcmp(var, "diff.suppress-blank-empty")) {
457 diff_suppress_blank_empty = git_config_bool(var, value);
458 return 0;
459 }
460
461 if (!strcmp(var, "diff.dirstat")) {
462 struct strbuf errmsg = STRBUF_INIT;
463 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
464 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
465 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
466 errmsg.buf);
467 strbuf_release(&errmsg);
468 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
469 return 0;
470 }
471
472 if (git_diff_heuristic_config(var, value, cb) < 0)
473 return -1;
474
475 return git_default_config(var, value, cb);
476 }
477
478 static char *quote_two(const char *one, const char *two)
479 {
480 int need_one = quote_c_style(one, NULL, NULL, 1);
481 int need_two = quote_c_style(two, NULL, NULL, 1);
482 struct strbuf res = STRBUF_INIT;
483
484 if (need_one + need_two) {
485 strbuf_addch(&res, '"');
486 quote_c_style(one, &res, NULL, 1);
487 quote_c_style(two, &res, NULL, 1);
488 strbuf_addch(&res, '"');
489 } else {
490 strbuf_addstr(&res, one);
491 strbuf_addstr(&res, two);
492 }
493 return strbuf_detach(&res, NULL);
494 }
495
496 static const char *external_diff(void)
497 {
498 static const char *external_diff_cmd = NULL;
499 static int done_preparing = 0;
500
501 if (done_preparing)
502 return external_diff_cmd;
503 external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
504 if (!external_diff_cmd)
505 external_diff_cmd = external_diff_cmd_cfg;
506 done_preparing = 1;
507 return external_diff_cmd;
508 }
509
510 /*
511 * Keep track of files used for diffing. Sometimes such an entry
512 * refers to a temporary file, sometimes to an existing file, and
513 * sometimes to "/dev/null".
514 */
515 static struct diff_tempfile {
516 /*
517 * filename external diff should read from, or NULL if this
518 * entry is currently not in use:
519 */
520 const char *name;
521
522 char hex[GIT_MAX_HEXSZ + 1];
523 char mode[10];
524
525 /*
526 * If this diff_tempfile instance refers to a temporary file,
527 * this tempfile object is used to manage its lifetime.
528 */
529 struct tempfile *tempfile;
530 } diff_temp[2];
531
532 struct emit_callback {
533 int color_diff;
534 unsigned ws_rule;
535 int blank_at_eof_in_preimage;
536 int blank_at_eof_in_postimage;
537 int lno_in_preimage;
538 int lno_in_postimage;
539 const char **label_path;
540 struct diff_words_data *diff_words;
541 struct diff_options *opt;
542 struct strbuf *header;
543 };
544
545 static int count_lines(const char *data, int size)
546 {
547 int count, ch, completely_empty = 1, nl_just_seen = 0;
548 count = 0;
549 while (0 < size--) {
550 ch = *data++;
551 if (ch == '\n') {
552 count++;
553 nl_just_seen = 1;
554 completely_empty = 0;
555 }
556 else {
557 nl_just_seen = 0;
558 completely_empty = 0;
559 }
560 }
561 if (completely_empty)
562 return 0;
563 if (!nl_just_seen)
564 count++; /* no trailing newline */
565 return count;
566 }
567
568 static int fill_mmfile(struct repository *r, mmfile_t *mf,
569 struct diff_filespec *one)
570 {
571 if (!DIFF_FILE_VALID(one)) {
572 mf->ptr = (char *)""; /* does not matter */
573 mf->size = 0;
574 return 0;
575 }
576 else if (diff_populate_filespec(r, one, 0))
577 return -1;
578
579 mf->ptr = one->data;
580 mf->size = one->size;
581 return 0;
582 }
583
584 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
585 static unsigned long diff_filespec_size(struct repository *r,
586 struct diff_filespec *one)
587 {
588 if (!DIFF_FILE_VALID(one))
589 return 0;
590 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
591 return one->size;
592 }
593
594 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
595 {
596 char *ptr = mf->ptr;
597 long size = mf->size;
598 int cnt = 0;
599
600 if (!size)
601 return cnt;
602 ptr += size - 1; /* pointing at the very end */
603 if (*ptr != '\n')
604 ; /* incomplete line */
605 else
606 ptr--; /* skip the last LF */
607 while (mf->ptr < ptr) {
608 char *prev_eol;
609 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
610 if (*prev_eol == '\n')
611 break;
612 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
613 break;
614 cnt++;
615 ptr = prev_eol - 1;
616 }
617 return cnt;
618 }
619
620 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
621 struct emit_callback *ecbdata)
622 {
623 int l1, l2, at;
624 unsigned ws_rule = ecbdata->ws_rule;
625 l1 = count_trailing_blank(mf1, ws_rule);
626 l2 = count_trailing_blank(mf2, ws_rule);
627 if (l2 <= l1) {
628 ecbdata->blank_at_eof_in_preimage = 0;
629 ecbdata->blank_at_eof_in_postimage = 0;
630 return;
631 }
632 at = count_lines(mf1->ptr, mf1->size);
633 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
634
635 at = count_lines(mf2->ptr, mf2->size);
636 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
637 }
638
639 static void emit_line_0(struct diff_options *o,
640 const char *set_sign, const char *set, unsigned reverse, const char *reset,
641 int first, const char *line, int len)
642 {
643 int has_trailing_newline, has_trailing_carriage_return;
644 int needs_reset = 0; /* at the end of the line */
645 FILE *file = o->file;
646
647 fputs(diff_line_prefix(o), file);
648
649 has_trailing_newline = (len > 0 && line[len-1] == '\n');
650 if (has_trailing_newline)
651 len--;
652
653 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
654 if (has_trailing_carriage_return)
655 len--;
656
657 if (!len && !first)
658 goto end_of_line;
659
660 if (reverse && want_color(o->use_color)) {
661 fputs(GIT_COLOR_REVERSE, file);
662 needs_reset = 1;
663 }
664
665 if (set_sign) {
666 fputs(set_sign, file);
667 needs_reset = 1;
668 }
669
670 if (first)
671 fputc(first, file);
672
673 if (!len)
674 goto end_of_line;
675
676 if (set) {
677 if (set_sign && set != set_sign)
678 fputs(reset, file);
679 fputs(set, file);
680 needs_reset = 1;
681 }
682 fwrite(line, len, 1, file);
683 needs_reset = 1; /* 'line' may contain color codes. */
684
685 end_of_line:
686 if (needs_reset)
687 fputs(reset, file);
688 if (has_trailing_carriage_return)
689 fputc('\r', file);
690 if (has_trailing_newline)
691 fputc('\n', file);
692 }
693
694 static void emit_line(struct diff_options *o, const char *set, const char *reset,
695 const char *line, int len)
696 {
697 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
698 }
699
700 enum diff_symbol {
701 DIFF_SYMBOL_BINARY_DIFF_HEADER,
702 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
703 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
704 DIFF_SYMBOL_BINARY_DIFF_BODY,
705 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
706 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
707 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
708 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
709 DIFF_SYMBOL_STATS_LINE,
710 DIFF_SYMBOL_WORD_DIFF,
711 DIFF_SYMBOL_STAT_SEP,
712 DIFF_SYMBOL_SUMMARY,
713 DIFF_SYMBOL_SUBMODULE_ADD,
714 DIFF_SYMBOL_SUBMODULE_DEL,
715 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
716 DIFF_SYMBOL_SUBMODULE_MODIFIED,
717 DIFF_SYMBOL_SUBMODULE_HEADER,
718 DIFF_SYMBOL_SUBMODULE_ERROR,
719 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
720 DIFF_SYMBOL_REWRITE_DIFF,
721 DIFF_SYMBOL_BINARY_FILES,
722 DIFF_SYMBOL_HEADER,
723 DIFF_SYMBOL_FILEPAIR_PLUS,
724 DIFF_SYMBOL_FILEPAIR_MINUS,
725 DIFF_SYMBOL_WORDS_PORCELAIN,
726 DIFF_SYMBOL_WORDS,
727 DIFF_SYMBOL_CONTEXT,
728 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
729 DIFF_SYMBOL_PLUS,
730 DIFF_SYMBOL_MINUS,
731 DIFF_SYMBOL_NO_LF_EOF,
732 DIFF_SYMBOL_CONTEXT_FRAGINFO,
733 DIFF_SYMBOL_CONTEXT_MARKER,
734 DIFF_SYMBOL_SEPARATOR
735 };
736 /*
737 * Flags for content lines:
738 * 0..12 are whitespace rules
739 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
740 * 16 is marking if the line is blank at EOF
741 */
742 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
743 #define DIFF_SYMBOL_MOVED_LINE (1<<17)
744 #define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
745 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
746 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
747
748 /*
749 * This struct is used when we need to buffer the output of the diff output.
750 *
751 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
752 * into the pre/post image file. This pointer could be a union with the
753 * line pointer. By storing an offset into the file instead of the literal line,
754 * we can decrease the memory footprint for the buffered output. At first we
755 * may want to only have indirection for the content lines, but we could also
756 * enhance the state for emitting prefabricated lines, e.g. the similarity
757 * score line or hunk/file headers would only need to store a number or path
758 * and then the output can be constructed later on depending on state.
759 */
760 struct emitted_diff_symbol {
761 const char *line;
762 int len;
763 int flags;
764 int indent_off; /* Offset to first non-whitespace character */
765 int indent_width; /* The visual width of the indentation */
766 enum diff_symbol s;
767 };
768 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
769
770 struct emitted_diff_symbols {
771 struct emitted_diff_symbol *buf;
772 int nr, alloc;
773 };
774 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
775
776 static void append_emitted_diff_symbol(struct diff_options *o,
777 struct emitted_diff_symbol *e)
778 {
779 struct emitted_diff_symbol *f;
780
781 ALLOC_GROW(o->emitted_symbols->buf,
782 o->emitted_symbols->nr + 1,
783 o->emitted_symbols->alloc);
784 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
785
786 memcpy(f, e, sizeof(struct emitted_diff_symbol));
787 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
788 }
789
790 struct moved_entry {
791 struct hashmap_entry ent;
792 const struct emitted_diff_symbol *es;
793 struct moved_entry *next_line;
794 };
795
796 struct moved_block {
797 struct moved_entry *match;
798 int wsd; /* The whitespace delta of this block */
799 };
800
801 static void moved_block_clear(struct moved_block *b)
802 {
803 memset(b, 0, sizeof(*b));
804 }
805
806 #define INDENT_BLANKLINE INT_MIN
807
808 static void fill_es_indent_data(struct emitted_diff_symbol *es)
809 {
810 unsigned int off = 0, i;
811 int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
812 const char *s = es->line;
813 const int len = es->len;
814
815 /* skip any \v \f \r at start of indentation */
816 while (s[off] == '\f' || s[off] == '\v' ||
817 (s[off] == '\r' && off < len - 1))
818 off++;
819
820 /* calculate the visual width of indentation */
821 while(1) {
822 if (s[off] == ' ') {
823 width++;
824 off++;
825 } else if (s[off] == '\t') {
826 width += tab_width - (width % tab_width);
827 while (s[++off] == '\t')
828 width += tab_width;
829 } else {
830 break;
831 }
832 }
833
834 /* check if this line is blank */
835 for (i = off; i < len; i++)
836 if (!isspace(s[i]))
837 break;
838
839 if (i == len) {
840 es->indent_width = INDENT_BLANKLINE;
841 es->indent_off = len;
842 } else {
843 es->indent_off = off;
844 es->indent_width = width;
845 }
846 }
847
848 static int compute_ws_delta(const struct emitted_diff_symbol *a,
849 const struct emitted_diff_symbol *b,
850 int *out)
851 {
852 int a_len = a->len,
853 b_len = b->len,
854 a_off = a->indent_off,
855 a_width = a->indent_width,
856 b_off = b->indent_off,
857 b_width = b->indent_width;
858 int delta;
859
860 if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
861 *out = INDENT_BLANKLINE;
862 return 1;
863 }
864
865 if (a->s == DIFF_SYMBOL_PLUS)
866 delta = a_width - b_width;
867 else
868 delta = b_width - a_width;
869
870 if (a_len - a_off != b_len - b_off ||
871 memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
872 return 0;
873
874 *out = delta;
875
876 return 1;
877 }
878
879 static int cmp_in_block_with_wsd(const struct diff_options *o,
880 const struct moved_entry *cur,
881 const struct moved_entry *match,
882 struct moved_block *pmb,
883 int n)
884 {
885 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
886 int al = cur->es->len, bl = match->es->len, cl = l->len;
887 const char *a = cur->es->line,
888 *b = match->es->line,
889 *c = l->line;
890 int a_off = cur->es->indent_off,
891 a_width = cur->es->indent_width,
892 c_off = l->indent_off,
893 c_width = l->indent_width;
894 int delta;
895
896 /*
897 * We need to check if 'cur' is equal to 'match'. As those
898 * are from the same (+/-) side, we do not need to adjust for
899 * indent changes. However these were found using fuzzy
900 * matching so we do have to check if they are equal. Here we
901 * just check the lengths. We delay calling memcmp() to check
902 * the contents until later as if the length comparison for a
903 * and c fails we can avoid the call all together.
904 */
905 if (al != bl)
906 return 1;
907
908 /* If 'l' and 'cur' are both blank then they match. */
909 if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
910 return 0;
911
912 /*
913 * The indent changes of the block are known and stored in pmb->wsd;
914 * however we need to check if the indent changes of the current line
915 * match those of the current block and that the text of 'l' and 'cur'
916 * after the indentation match.
917 */
918 if (cur->es->s == DIFF_SYMBOL_PLUS)
919 delta = a_width - c_width;
920 else
921 delta = c_width - a_width;
922
923 /*
924 * If the previous lines of this block were all blank then set its
925 * whitespace delta.
926 */
927 if (pmb->wsd == INDENT_BLANKLINE)
928 pmb->wsd = delta;
929
930 return !(delta == pmb->wsd && al - a_off == cl - c_off &&
931 !memcmp(a, b, al) && !
932 memcmp(a + a_off, c + c_off, al - a_off));
933 }
934
935 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
936 const void *entry,
937 const void *entry_or_key,
938 const void *keydata)
939 {
940 const struct diff_options *diffopt = hashmap_cmp_fn_data;
941 const struct moved_entry *a = entry;
942 const struct moved_entry *b = entry_or_key;
943 unsigned flags = diffopt->color_moved_ws_handling
944 & XDF_WHITESPACE_FLAGS;
945
946 if (diffopt->color_moved_ws_handling &
947 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
948 /*
949 * As there is not specific white space config given,
950 * we'd need to check for a new block, so ignore all
951 * white space. The setup of the white space
952 * configuration for the next block is done else where
953 */
954 flags |= XDF_IGNORE_WHITESPACE;
955
956 return !xdiff_compare_lines(a->es->line, a->es->len,
957 b->es->line, b->es->len,
958 flags);
959 }
960
961 static struct moved_entry *prepare_entry(struct diff_options *o,
962 int line_no)
963 {
964 struct moved_entry *ret = xmalloc(sizeof(*ret));
965 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
966 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
967
968 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
969 ret->es = l;
970 ret->next_line = NULL;
971
972 return ret;
973 }
974
975 static void add_lines_to_move_detection(struct diff_options *o,
976 struct hashmap *add_lines,
977 struct hashmap *del_lines)
978 {
979 struct moved_entry *prev_line = NULL;
980
981 int n;
982 for (n = 0; n < o->emitted_symbols->nr; n++) {
983 struct hashmap *hm;
984 struct moved_entry *key;
985
986 switch (o->emitted_symbols->buf[n].s) {
987 case DIFF_SYMBOL_PLUS:
988 hm = add_lines;
989 break;
990 case DIFF_SYMBOL_MINUS:
991 hm = del_lines;
992 break;
993 default:
994 prev_line = NULL;
995 continue;
996 }
997
998 if (o->color_moved_ws_handling &
999 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1000 fill_es_indent_data(&o->emitted_symbols->buf[n]);
1001 key = prepare_entry(o, n);
1002 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1003 prev_line->next_line = key;
1004
1005 hashmap_add(hm, key);
1006 prev_line = key;
1007 }
1008 }
1009
1010 static void pmb_advance_or_null(struct diff_options *o,
1011 struct moved_entry *match,
1012 struct hashmap *hm,
1013 struct moved_block *pmb,
1014 int pmb_nr)
1015 {
1016 int i;
1017 for (i = 0; i < pmb_nr; i++) {
1018 struct moved_entry *prev = pmb[i].match;
1019 struct moved_entry *cur = (prev && prev->next_line) ?
1020 prev->next_line : NULL;
1021 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1022 pmb[i].match = cur;
1023 } else {
1024 pmb[i].match = NULL;
1025 }
1026 }
1027 }
1028
1029 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1030 struct moved_entry *match,
1031 struct hashmap *hm,
1032 struct moved_block *pmb,
1033 int pmb_nr, int n)
1034 {
1035 int i;
1036 char *got_match = xcalloc(1, pmb_nr);
1037
1038 for (; match; match = hashmap_get_next(hm, match)) {
1039 for (i = 0; i < pmb_nr; i++) {
1040 struct moved_entry *prev = pmb[i].match;
1041 struct moved_entry *cur = (prev && prev->next_line) ?
1042 prev->next_line : NULL;
1043 if (!cur)
1044 continue;
1045 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1046 got_match[i] |= 1;
1047 }
1048 }
1049
1050 for (i = 0; i < pmb_nr; i++) {
1051 if (got_match[i]) {
1052 /* Advance to the next line */
1053 pmb[i].match = pmb[i].match->next_line;
1054 } else {
1055 moved_block_clear(&pmb[i]);
1056 }
1057 }
1058
1059 free(got_match);
1060 }
1061
1062 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1063 int pmb_nr)
1064 {
1065 int lp, rp;
1066
1067 /* Shrink the set of potential block to the remaining running */
1068 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1069 while (lp < pmb_nr && pmb[lp].match)
1070 lp++;
1071 /* lp points at the first NULL now */
1072
1073 while (rp > -1 && !pmb[rp].match)
1074 rp--;
1075 /* rp points at the last non-NULL */
1076
1077 if (lp < pmb_nr && rp > -1 && lp < rp) {
1078 pmb[lp] = pmb[rp];
1079 memset(&pmb[rp], 0, sizeof(pmb[rp]));
1080 rp--;
1081 lp++;
1082 }
1083 }
1084
1085 /* Remember the number of running sets */
1086 return rp + 1;
1087 }
1088
1089 /*
1090 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1091 *
1092 * Otherwise, if the last block has fewer alphanumeric characters than
1093 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1094 * that block.
1095 *
1096 * The last block consists of the (n - block_length)'th line up to but not
1097 * including the nth line.
1098 *
1099 * Returns 0 if the last block is empty or is unset by this function, non zero
1100 * otherwise.
1101 *
1102 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1103 * Think of a way to unify them.
1104 */
1105 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1106 {
1107 int i, alnum_count = 0;
1108 if (o->color_moved == COLOR_MOVED_PLAIN)
1109 return block_length;
1110 for (i = 1; i < block_length + 1; i++) {
1111 const char *c = o->emitted_symbols->buf[n - i].line;
1112 for (; *c; c++) {
1113 if (!isalnum(*c))
1114 continue;
1115 alnum_count++;
1116 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1117 return 1;
1118 }
1119 }
1120 for (i = 1; i < block_length + 1; i++)
1121 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1122 return 0;
1123 }
1124
1125 /* Find blocks of moved code, delegate actual coloring decision to helper */
1126 static void mark_color_as_moved(struct diff_options *o,
1127 struct hashmap *add_lines,
1128 struct hashmap *del_lines)
1129 {
1130 struct moved_block *pmb = NULL; /* potentially moved blocks */
1131 int pmb_nr = 0, pmb_alloc = 0;
1132 int n, flipped_block = 0, block_length = 0;
1133
1134
1135 for (n = 0; n < o->emitted_symbols->nr; n++) {
1136 struct hashmap *hm = NULL;
1137 struct moved_entry *key;
1138 struct moved_entry *match = NULL;
1139 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1140 enum diff_symbol last_symbol = 0;
1141
1142 switch (l->s) {
1143 case DIFF_SYMBOL_PLUS:
1144 hm = del_lines;
1145 key = prepare_entry(o, n);
1146 match = hashmap_get(hm, key, NULL);
1147 free(key);
1148 break;
1149 case DIFF_SYMBOL_MINUS:
1150 hm = add_lines;
1151 key = prepare_entry(o, n);
1152 match = hashmap_get(hm, key, NULL);
1153 free(key);
1154 break;
1155 default:
1156 flipped_block = 0;
1157 }
1158
1159 if (!match) {
1160 int i;
1161
1162 adjust_last_block(o, n, block_length);
1163 for(i = 0; i < pmb_nr; i++)
1164 moved_block_clear(&pmb[i]);
1165 pmb_nr = 0;
1166 block_length = 0;
1167 flipped_block = 0;
1168 last_symbol = l->s;
1169 continue;
1170 }
1171
1172 if (o->color_moved == COLOR_MOVED_PLAIN) {
1173 last_symbol = l->s;
1174 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1175 continue;
1176 }
1177
1178 if (o->color_moved_ws_handling &
1179 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1180 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1181 else
1182 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1183
1184 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1185
1186 if (pmb_nr == 0) {
1187 /*
1188 * The current line is the start of a new block.
1189 * Setup the set of potential blocks.
1190 */
1191 for (; match; match = hashmap_get_next(hm, match)) {
1192 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1193 if (o->color_moved_ws_handling &
1194 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1195 if (compute_ws_delta(l, match->es,
1196 &pmb[pmb_nr].wsd))
1197 pmb[pmb_nr++].match = match;
1198 } else {
1199 pmb[pmb_nr].wsd = 0;
1200 pmb[pmb_nr++].match = match;
1201 }
1202 }
1203
1204 if (adjust_last_block(o, n, block_length) &&
1205 pmb_nr && last_symbol != l->s)
1206 flipped_block = (flipped_block + 1) % 2;
1207 else
1208 flipped_block = 0;
1209
1210 block_length = 0;
1211 }
1212
1213 if (pmb_nr) {
1214 block_length++;
1215 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1216 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1217 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1218 }
1219 last_symbol = l->s;
1220 }
1221 adjust_last_block(o, n, block_length);
1222
1223 for(n = 0; n < pmb_nr; n++)
1224 moved_block_clear(&pmb[n]);
1225 free(pmb);
1226 }
1227
1228 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1229 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1230 static void dim_moved_lines(struct diff_options *o)
1231 {
1232 int n;
1233 for (n = 0; n < o->emitted_symbols->nr; n++) {
1234 struct emitted_diff_symbol *prev = (n != 0) ?
1235 &o->emitted_symbols->buf[n - 1] : NULL;
1236 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1237 struct emitted_diff_symbol *next =
1238 (n < o->emitted_symbols->nr - 1) ?
1239 &o->emitted_symbols->buf[n + 1] : NULL;
1240
1241 /* Not a plus or minus line? */
1242 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1243 continue;
1244
1245 /* Not a moved line? */
1246 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1247 continue;
1248
1249 /*
1250 * If prev or next are not a plus or minus line,
1251 * pretend they don't exist
1252 */
1253 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1254 prev->s != DIFF_SYMBOL_MINUS)
1255 prev = NULL;
1256 if (next && next->s != DIFF_SYMBOL_PLUS &&
1257 next->s != DIFF_SYMBOL_MINUS)
1258 next = NULL;
1259
1260 /* Inside a block? */
1261 if ((prev &&
1262 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1263 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1264 (next &&
1265 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1266 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1267 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1268 continue;
1269 }
1270
1271 /* Check if we are at an interesting bound: */
1272 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1273 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1274 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1275 continue;
1276 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1277 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1278 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1279 continue;
1280
1281 /*
1282 * The boundary to prev and next are not interesting,
1283 * so this line is not interesting as a whole
1284 */
1285 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1286 }
1287 }
1288
1289 static void emit_line_ws_markup(struct diff_options *o,
1290 const char *set_sign, const char *set,
1291 const char *reset,
1292 int sign_index, const char *line, int len,
1293 unsigned ws_rule, int blank_at_eof)
1294 {
1295 const char *ws = NULL;
1296 int sign = o->output_indicators[sign_index];
1297
1298 if (o->ws_error_highlight & ws_rule) {
1299 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1300 if (!*ws)
1301 ws = NULL;
1302 }
1303
1304 if (!ws && !set_sign)
1305 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1306 else if (!ws) {
1307 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1308 } else if (blank_at_eof)
1309 /* Blank line at EOF - paint '+' as well */
1310 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1311 else {
1312 /* Emit just the prefix, then the rest. */
1313 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1314 sign, "", 0);
1315 ws_check_emit(line, len, ws_rule,
1316 o->file, set, reset, ws);
1317 }
1318 }
1319
1320 static void emit_diff_symbol_from_struct(struct diff_options *o,
1321 struct emitted_diff_symbol *eds)
1322 {
1323 static const char *nneof = " No newline at end of file\n";
1324 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1325 struct strbuf sb = STRBUF_INIT;
1326
1327 enum diff_symbol s = eds->s;
1328 const char *line = eds->line;
1329 int len = eds->len;
1330 unsigned flags = eds->flags;
1331
1332 switch (s) {
1333 case DIFF_SYMBOL_NO_LF_EOF:
1334 context = diff_get_color_opt(o, DIFF_CONTEXT);
1335 reset = diff_get_color_opt(o, DIFF_RESET);
1336 putc('\n', o->file);
1337 emit_line_0(o, context, NULL, 0, reset, '\\',
1338 nneof, strlen(nneof));
1339 break;
1340 case DIFF_SYMBOL_SUBMODULE_HEADER:
1341 case DIFF_SYMBOL_SUBMODULE_ERROR:
1342 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1343 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1344 case DIFF_SYMBOL_SUMMARY:
1345 case DIFF_SYMBOL_STATS_LINE:
1346 case DIFF_SYMBOL_BINARY_DIFF_BODY:
1347 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1348 emit_line(o, "", "", line, len);
1349 break;
1350 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1351 case DIFF_SYMBOL_CONTEXT_MARKER:
1352 context = diff_get_color_opt(o, DIFF_CONTEXT);
1353 reset = diff_get_color_opt(o, DIFF_RESET);
1354 emit_line(o, context, reset, line, len);
1355 break;
1356 case DIFF_SYMBOL_SEPARATOR:
1357 fprintf(o->file, "%s%c",
1358 diff_line_prefix(o),
1359 o->line_termination);
1360 break;
1361 case DIFF_SYMBOL_CONTEXT:
1362 set = diff_get_color_opt(o, DIFF_CONTEXT);
1363 reset = diff_get_color_opt(o, DIFF_RESET);
1364 set_sign = NULL;
1365 if (o->flags.dual_color_diffed_diffs) {
1366 char c = !len ? 0 : line[0];
1367
1368 if (c == '+')
1369 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1370 else if (c == '@')
1371 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1372 else if (c == '-')
1373 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1374 }
1375 emit_line_ws_markup(o, set_sign, set, reset,
1376 OUTPUT_INDICATOR_CONTEXT, line, len,
1377 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1378 break;
1379 case DIFF_SYMBOL_PLUS:
1380 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1381 DIFF_SYMBOL_MOVED_LINE_ALT |
1382 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1383 case DIFF_SYMBOL_MOVED_LINE |
1384 DIFF_SYMBOL_MOVED_LINE_ALT |
1385 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1386 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1387 break;
1388 case DIFF_SYMBOL_MOVED_LINE |
1389 DIFF_SYMBOL_MOVED_LINE_ALT:
1390 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1391 break;
1392 case DIFF_SYMBOL_MOVED_LINE |
1393 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1394 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1395 break;
1396 case DIFF_SYMBOL_MOVED_LINE:
1397 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1398 break;
1399 default:
1400 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1401 }
1402 reset = diff_get_color_opt(o, DIFF_RESET);
1403 if (!o->flags.dual_color_diffed_diffs)
1404 set_sign = NULL;
1405 else {
1406 char c = !len ? 0 : line[0];
1407
1408 set_sign = set;
1409 if (c == '-')
1410 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1411 else if (c == '@')
1412 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1413 else if (c == '+')
1414 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1415 else
1416 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1417 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1418 }
1419 emit_line_ws_markup(o, set_sign, set, reset,
1420 OUTPUT_INDICATOR_NEW, line, len,
1421 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1422 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1423 break;
1424 case DIFF_SYMBOL_MINUS:
1425 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1426 DIFF_SYMBOL_MOVED_LINE_ALT |
1427 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1428 case DIFF_SYMBOL_MOVED_LINE |
1429 DIFF_SYMBOL_MOVED_LINE_ALT |
1430 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1431 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1432 break;
1433 case DIFF_SYMBOL_MOVED_LINE |
1434 DIFF_SYMBOL_MOVED_LINE_ALT:
1435 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1436 break;
1437 case DIFF_SYMBOL_MOVED_LINE |
1438 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1439 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1440 break;
1441 case DIFF_SYMBOL_MOVED_LINE:
1442 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1443 break;
1444 default:
1445 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1446 }
1447 reset = diff_get_color_opt(o, DIFF_RESET);
1448 if (!o->flags.dual_color_diffed_diffs)
1449 set_sign = NULL;
1450 else {
1451 char c = !len ? 0 : line[0];
1452
1453 set_sign = set;
1454 if (c == '+')
1455 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1456 else if (c == '@')
1457 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1458 else if (c == '-')
1459 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1460 else
1461 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1462 }
1463 emit_line_ws_markup(o, set_sign, set, reset,
1464 OUTPUT_INDICATOR_OLD, line, len,
1465 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1466 break;
1467 case DIFF_SYMBOL_WORDS_PORCELAIN:
1468 context = diff_get_color_opt(o, DIFF_CONTEXT);
1469 reset = diff_get_color_opt(o, DIFF_RESET);
1470 emit_line(o, context, reset, line, len);
1471 fputs("~\n", o->file);
1472 break;
1473 case DIFF_SYMBOL_WORDS:
1474 context = diff_get_color_opt(o, DIFF_CONTEXT);
1475 reset = diff_get_color_opt(o, DIFF_RESET);
1476 /*
1477 * Skip the prefix character, if any. With
1478 * diff_suppress_blank_empty, there may be
1479 * none.
1480 */
1481 if (line[0] != '\n') {
1482 line++;
1483 len--;
1484 }
1485 emit_line(o, context, reset, line, len);
1486 break;
1487 case DIFF_SYMBOL_FILEPAIR_PLUS:
1488 meta = diff_get_color_opt(o, DIFF_METAINFO);
1489 reset = diff_get_color_opt(o, DIFF_RESET);
1490 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1491 line, reset,
1492 strchr(line, ' ') ? "\t" : "");
1493 break;
1494 case DIFF_SYMBOL_FILEPAIR_MINUS:
1495 meta = diff_get_color_opt(o, DIFF_METAINFO);
1496 reset = diff_get_color_opt(o, DIFF_RESET);
1497 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1498 line, reset,
1499 strchr(line, ' ') ? "\t" : "");
1500 break;
1501 case DIFF_SYMBOL_BINARY_FILES:
1502 case DIFF_SYMBOL_HEADER:
1503 fprintf(o->file, "%s", line);
1504 break;
1505 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1506 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1507 break;
1508 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1509 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1510 break;
1511 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1512 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1513 break;
1514 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1515 fputs(diff_line_prefix(o), o->file);
1516 fputc('\n', o->file);
1517 break;
1518 case DIFF_SYMBOL_REWRITE_DIFF:
1519 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1520 reset = diff_get_color_opt(o, DIFF_RESET);
1521 emit_line(o, fraginfo, reset, line, len);
1522 break;
1523 case DIFF_SYMBOL_SUBMODULE_ADD:
1524 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1525 reset = diff_get_color_opt(o, DIFF_RESET);
1526 emit_line(o, set, reset, line, len);
1527 break;
1528 case DIFF_SYMBOL_SUBMODULE_DEL:
1529 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1530 reset = diff_get_color_opt(o, DIFF_RESET);
1531 emit_line(o, set, reset, line, len);
1532 break;
1533 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1534 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1535 diff_line_prefix(o), line);
1536 break;
1537 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1538 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1539 diff_line_prefix(o), line);
1540 break;
1541 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1542 emit_line(o, "", "", " 0 files changed\n",
1543 strlen(" 0 files changed\n"));
1544 break;
1545 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1546 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1547 break;
1548 case DIFF_SYMBOL_WORD_DIFF:
1549 fprintf(o->file, "%.*s", len, line);
1550 break;
1551 case DIFF_SYMBOL_STAT_SEP:
1552 fputs(o->stat_sep, o->file);
1553 break;
1554 default:
1555 BUG("unknown diff symbol");
1556 }
1557 strbuf_release(&sb);
1558 }
1559
1560 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1561 const char *line, int len, unsigned flags)
1562 {
1563 struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1564
1565 if (o->emitted_symbols)
1566 append_emitted_diff_symbol(o, &e);
1567 else
1568 emit_diff_symbol_from_struct(o, &e);
1569 }
1570
1571 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1572 {
1573 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1574 }
1575
1576 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1577 {
1578 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1579 }
1580
1581 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1582 {
1583 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1584 path, strlen(path), 0);
1585 }
1586
1587 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1588 {
1589 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1590 path, strlen(path), 0);
1591 }
1592
1593 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1594 {
1595 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1596 header, strlen(header), 0);
1597 }
1598
1599 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1600 {
1601 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1602 }
1603
1604 void diff_emit_submodule_pipethrough(struct diff_options *o,
1605 const char *line, int len)
1606 {
1607 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1608 }
1609
1610 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1611 {
1612 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1613 ecbdata->blank_at_eof_in_preimage &&
1614 ecbdata->blank_at_eof_in_postimage &&
1615 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1616 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1617 return 0;
1618 return ws_blank_line(line, len, ecbdata->ws_rule);
1619 }
1620
1621 static void emit_add_line(struct emit_callback *ecbdata,
1622 const char *line, int len)
1623 {
1624 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1625 if (new_blank_line_at_eof(ecbdata, line, len))
1626 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1627
1628 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1629 }
1630
1631 static void emit_del_line(struct emit_callback *ecbdata,
1632 const char *line, int len)
1633 {
1634 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1635 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1636 }
1637
1638 static void emit_context_line(struct emit_callback *ecbdata,
1639 const char *line, int len)
1640 {
1641 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1642 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1643 }
1644
1645 static void emit_hunk_header(struct emit_callback *ecbdata,
1646 const char *line, int len)
1647 {
1648 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1649 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1650 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1651 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1652 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1653 static const char atat[2] = { '@', '@' };
1654 const char *cp, *ep;
1655 struct strbuf msgbuf = STRBUF_INIT;
1656 int org_len = len;
1657 int i = 1;
1658
1659 /*
1660 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1661 * it always is at least 10 bytes long.
1662 */
1663 if (len < 10 ||
1664 memcmp(line, atat, 2) ||
1665 !(ep = memmem(line + 2, len - 2, atat, 2))) {
1666 emit_diff_symbol(ecbdata->opt,
1667 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1668 return;
1669 }
1670 ep += 2; /* skip over @@ */
1671
1672 /* The hunk header in fraginfo color */
1673 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1674 strbuf_addstr(&msgbuf, reverse);
1675 strbuf_addstr(&msgbuf, frag);
1676 if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1677 strbuf_add(&msgbuf, atat, sizeof(atat));
1678 else
1679 strbuf_add(&msgbuf, line, ep - line);
1680 strbuf_addstr(&msgbuf, reset);
1681
1682 /*
1683 * trailing "\r\n"
1684 */
1685 for ( ; i < 3; i++)
1686 if (line[len - i] == '\r' || line[len - i] == '\n')
1687 len--;
1688
1689 /* blank before the func header */
1690 for (cp = ep; ep - line < len; ep++)
1691 if (*ep != ' ' && *ep != '\t')
1692 break;
1693 if (ep != cp) {
1694 strbuf_addstr(&msgbuf, context);
1695 strbuf_add(&msgbuf, cp, ep - cp);
1696 strbuf_addstr(&msgbuf, reset);
1697 }
1698
1699 if (ep < line + len) {
1700 strbuf_addstr(&msgbuf, func);
1701 strbuf_add(&msgbuf, ep, line + len - ep);
1702 strbuf_addstr(&msgbuf, reset);
1703 }
1704
1705 strbuf_add(&msgbuf, line + len, org_len - len);
1706 strbuf_complete_line(&msgbuf);
1707 emit_diff_symbol(ecbdata->opt,
1708 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1709 strbuf_release(&msgbuf);
1710 }
1711
1712 static struct diff_tempfile *claim_diff_tempfile(void)
1713 {
1714 int i;
1715 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1716 if (!diff_temp[i].name)
1717 return diff_temp + i;
1718 BUG("diff is failing to clean up its tempfiles");
1719 }
1720
1721 static void remove_tempfile(void)
1722 {
1723 int i;
1724 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1725 if (is_tempfile_active(diff_temp[i].tempfile))
1726 delete_tempfile(&diff_temp[i].tempfile);
1727 diff_temp[i].name = NULL;
1728 }
1729 }
1730
1731 static void add_line_count(struct strbuf *out, int count)
1732 {
1733 switch (count) {
1734 case 0:
1735 strbuf_addstr(out, "0,0");
1736 break;
1737 case 1:
1738 strbuf_addstr(out, "1");
1739 break;
1740 default:
1741 strbuf_addf(out, "1,%d", count);
1742 break;
1743 }
1744 }
1745
1746 static void emit_rewrite_lines(struct emit_callback *ecb,
1747 int prefix, const char *data, int size)
1748 {
1749 const char *endp = NULL;
1750
1751 while (0 < size) {
1752 int len;
1753
1754 endp = memchr(data, '\n', size);
1755 len = endp ? (endp - data + 1) : size;
1756 if (prefix != '+') {
1757 ecb->lno_in_preimage++;
1758 emit_del_line(ecb, data, len);
1759 } else {
1760 ecb->lno_in_postimage++;
1761 emit_add_line(ecb, data, len);
1762 }
1763 size -= len;
1764 data += len;
1765 }
1766 if (!endp)
1767 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1768 }
1769
1770 static void emit_rewrite_diff(const char *name_a,
1771 const char *name_b,
1772 struct diff_filespec *one,
1773 struct diff_filespec *two,
1774 struct userdiff_driver *textconv_one,
1775 struct userdiff_driver *textconv_two,
1776 struct diff_options *o)
1777 {
1778 int lc_a, lc_b;
1779 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1780 const char *a_prefix, *b_prefix;
1781 char *data_one, *data_two;
1782 size_t size_one, size_two;
1783 struct emit_callback ecbdata;
1784 struct strbuf out = STRBUF_INIT;
1785
1786 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1787 a_prefix = o->b_prefix;
1788 b_prefix = o->a_prefix;
1789 } else {
1790 a_prefix = o->a_prefix;
1791 b_prefix = o->b_prefix;
1792 }
1793
1794 name_a += (*name_a == '/');
1795 name_b += (*name_b == '/');
1796
1797 strbuf_reset(&a_name);
1798 strbuf_reset(&b_name);
1799 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1800 quote_two_c_style(&b_name, b_prefix, name_b, 0);
1801
1802 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1803 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1804
1805 memset(&ecbdata, 0, sizeof(ecbdata));
1806 ecbdata.color_diff = want_color(o->use_color);
1807 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1808 ecbdata.opt = o;
1809 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1810 mmfile_t mf1, mf2;
1811 mf1.ptr = (char *)data_one;
1812 mf2.ptr = (char *)data_two;
1813 mf1.size = size_one;
1814 mf2.size = size_two;
1815 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1816 }
1817 ecbdata.lno_in_preimage = 1;
1818 ecbdata.lno_in_postimage = 1;
1819
1820 lc_a = count_lines(data_one, size_one);
1821 lc_b = count_lines(data_two, size_two);
1822
1823 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1824 a_name.buf, a_name.len, 0);
1825 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1826 b_name.buf, b_name.len, 0);
1827
1828 strbuf_addstr(&out, "@@ -");
1829 if (!o->irreversible_delete)
1830 add_line_count(&out, lc_a);
1831 else
1832 strbuf_addstr(&out, "?,?");
1833 strbuf_addstr(&out, " +");
1834 add_line_count(&out, lc_b);
1835 strbuf_addstr(&out, " @@\n");
1836 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1837 strbuf_release(&out);
1838
1839 if (lc_a && !o->irreversible_delete)
1840 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1841 if (lc_b)
1842 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1843 if (textconv_one)
1844 free((char *)data_one);
1845 if (textconv_two)
1846 free((char *)data_two);
1847 }
1848
1849 struct diff_words_buffer {
1850 mmfile_t text;
1851 unsigned long alloc;
1852 struct diff_words_orig {
1853 const char *begin, *end;
1854 } *orig;
1855 int orig_nr, orig_alloc;
1856 };
1857
1858 static void diff_words_append(char *line, unsigned long len,
1859 struct diff_words_buffer *buffer)
1860 {
1861 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1862 line++;
1863 len--;
1864 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1865 buffer->text.size += len;
1866 buffer->text.ptr[buffer->text.size] = '\0';
1867 }
1868
1869 struct diff_words_style_elem {
1870 const char *prefix;
1871 const char *suffix;
1872 const char *color; /* NULL; filled in by the setup code if
1873 * color is enabled */
1874 };
1875
1876 struct diff_words_style {
1877 enum diff_words_type type;
1878 struct diff_words_style_elem new_word, old_word, ctx;
1879 const char *newline;
1880 };
1881
1882 static struct diff_words_style diff_words_styles[] = {
1883 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1884 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1885 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1886 };
1887
1888 struct diff_words_data {
1889 struct diff_words_buffer minus, plus;
1890 const char *current_plus;
1891 int last_minus;
1892 struct diff_options *opt;
1893 regex_t *word_regex;
1894 enum diff_words_type type;
1895 struct diff_words_style *style;
1896 };
1897
1898 static int fn_out_diff_words_write_helper(struct diff_options *o,
1899 struct diff_words_style_elem *st_el,
1900 const char *newline,
1901 size_t count, const char *buf)
1902 {
1903 int print = 0;
1904 struct strbuf sb = STRBUF_INIT;
1905
1906 while (count) {
1907 char *p = memchr(buf, '\n', count);
1908 if (print)
1909 strbuf_addstr(&sb, diff_line_prefix(o));
1910
1911 if (p != buf) {
1912 const char *reset = st_el->color && *st_el->color ?
1913 GIT_COLOR_RESET : NULL;
1914 if (st_el->color && *st_el->color)
1915 strbuf_addstr(&sb, st_el->color);
1916 strbuf_addstr(&sb, st_el->prefix);
1917 strbuf_add(&sb, buf, p ? p - buf : count);
1918 strbuf_addstr(&sb, st_el->suffix);
1919 if (reset)
1920 strbuf_addstr(&sb, reset);
1921 }
1922 if (!p)
1923 goto out;
1924
1925 strbuf_addstr(&sb, newline);
1926 count -= p + 1 - buf;
1927 buf = p + 1;
1928 print = 1;
1929 if (count) {
1930 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1931 sb.buf, sb.len, 0);
1932 strbuf_reset(&sb);
1933 }
1934 }
1935
1936 out:
1937 if (sb.len)
1938 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1939 sb.buf, sb.len, 0);
1940 strbuf_release(&sb);
1941 return 0;
1942 }
1943
1944 /*
1945 * '--color-words' algorithm can be described as:
1946 *
1947 * 1. collect the minus/plus lines of a diff hunk, divided into
1948 * minus-lines and plus-lines;
1949 *
1950 * 2. break both minus-lines and plus-lines into words and
1951 * place them into two mmfile_t with one word for each line;
1952 *
1953 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1954 *
1955 * And for the common parts of the both file, we output the plus side text.
1956 * diff_words->current_plus is used to trace the current position of the plus file
1957 * which printed. diff_words->last_minus is used to trace the last minus word
1958 * printed.
1959 *
1960 * For '--graph' to work with '--color-words', we need to output the graph prefix
1961 * on each line of color words output. Generally, there are two conditions on
1962 * which we should output the prefix.
1963 *
1964 * 1. diff_words->last_minus == 0 &&
1965 * diff_words->current_plus == diff_words->plus.text.ptr
1966 *
1967 * that is: the plus text must start as a new line, and if there is no minus
1968 * word printed, a graph prefix must be printed.
1969 *
1970 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1971 * *(diff_words->current_plus - 1) == '\n'
1972 *
1973 * that is: a graph prefix must be printed following a '\n'
1974 */
1975 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1976 {
1977 if ((diff_words->last_minus == 0 &&
1978 diff_words->current_plus == diff_words->plus.text.ptr) ||
1979 (diff_words->current_plus > diff_words->plus.text.ptr &&
1980 *(diff_words->current_plus - 1) == '\n')) {
1981 return 1;
1982 } else {
1983 return 0;
1984 }
1985 }
1986
1987 static void fn_out_diff_words_aux(void *priv,
1988 long minus_first, long minus_len,
1989 long plus_first, long plus_len,
1990 const char *func, long funclen)
1991 {
1992 struct diff_words_data *diff_words = priv;
1993 struct diff_words_style *style = diff_words->style;
1994 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1995 struct diff_options *opt = diff_words->opt;
1996 const char *line_prefix;
1997
1998 assert(opt);
1999 line_prefix = diff_line_prefix(opt);
2000
2001 /* POSIX requires that first be decremented by one if len == 0... */
2002 if (minus_len) {
2003 minus_begin = diff_words->minus.orig[minus_first].begin;
2004 minus_end =
2005 diff_words->minus.orig[minus_first + minus_len - 1].end;
2006 } else
2007 minus_begin = minus_end =
2008 diff_words->minus.orig[minus_first].end;
2009
2010 if (plus_len) {
2011 plus_begin = diff_words->plus.orig[plus_first].begin;
2012 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2013 } else
2014 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2015
2016 if (color_words_output_graph_prefix(diff_words)) {
2017 fputs(line_prefix, diff_words->opt->file);
2018 }
2019 if (diff_words->current_plus != plus_begin) {
2020 fn_out_diff_words_write_helper(diff_words->opt,
2021 &style->ctx, style->newline,
2022 plus_begin - diff_words->current_plus,
2023 diff_words->current_plus);
2024 }
2025 if (minus_begin != minus_end) {
2026 fn_out_diff_words_write_helper(diff_words->opt,
2027 &style->old_word, style->newline,
2028 minus_end - minus_begin, minus_begin);
2029 }
2030 if (plus_begin != plus_end) {
2031 fn_out_diff_words_write_helper(diff_words->opt,
2032 &style->new_word, style->newline,
2033 plus_end - plus_begin, plus_begin);
2034 }
2035
2036 diff_words->current_plus = plus_end;
2037 diff_words->last_minus = minus_first;
2038 }
2039
2040 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2041 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2042 int *begin, int *end)
2043 {
2044 if (word_regex && *begin < buffer->size) {
2045 regmatch_t match[1];
2046 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2047 buffer->size - *begin, 1, match, 0)) {
2048 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2049 '\n', match[0].rm_eo - match[0].rm_so);
2050 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2051 *begin += match[0].rm_so;
2052 return *begin >= *end;
2053 }
2054 return -1;
2055 }
2056
2057 /* find the next word */
2058 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2059 (*begin)++;
2060 if (*begin >= buffer->size)
2061 return -1;
2062
2063 /* find the end of the word */
2064 *end = *begin + 1;
2065 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2066 (*end)++;
2067
2068 return 0;
2069 }
2070
2071 /*
2072 * This function splits the words in buffer->text, stores the list with
2073 * newline separator into out, and saves the offsets of the original words
2074 * in buffer->orig.
2075 */
2076 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2077 regex_t *word_regex)
2078 {
2079 int i, j;
2080 long alloc = 0;
2081
2082 out->size = 0;
2083 out->ptr = NULL;
2084
2085 /* fake an empty "0th" word */
2086 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2087 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2088 buffer->orig_nr = 1;
2089
2090 for (i = 0; i < buffer->text.size; i++) {
2091 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2092 return;
2093
2094 /* store original boundaries */
2095 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2096 buffer->orig_alloc);
2097 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2098 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2099 buffer->orig_nr++;
2100
2101 /* store one word */
2102 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2103 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2104 out->ptr[out->size + j - i] = '\n';
2105 out->size += j - i + 1;
2106
2107 i = j - 1;
2108 }
2109 }
2110
2111 /* this executes the word diff on the accumulated buffers */
2112 static void diff_words_show(struct diff_words_data *diff_words)
2113 {
2114 xpparam_t xpp;
2115 xdemitconf_t xecfg;
2116 mmfile_t minus, plus;
2117 struct diff_words_style *style = diff_words->style;
2118
2119 struct diff_options *opt = diff_words->opt;
2120 const char *line_prefix;
2121
2122 assert(opt);
2123 line_prefix = diff_line_prefix(opt);
2124
2125 /* special case: only removal */
2126 if (!diff_words->plus.text.size) {
2127 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2128 line_prefix, strlen(line_prefix), 0);
2129 fn_out_diff_words_write_helper(diff_words->opt,
2130 &style->old_word, style->newline,
2131 diff_words->minus.text.size,
2132 diff_words->minus.text.ptr);
2133 diff_words->minus.text.size = 0;
2134 return;
2135 }
2136
2137 diff_words->current_plus = diff_words->plus.text.ptr;
2138 diff_words->last_minus = 0;
2139
2140 memset(&xpp, 0, sizeof(xpp));
2141 memset(&xecfg, 0, sizeof(xecfg));
2142 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2143 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2144 xpp.flags = 0;
2145 /* as only the hunk header will be parsed, we need a 0-context */
2146 xecfg.ctxlen = 0;
2147 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2148 diff_words, &xpp, &xecfg))
2149 die("unable to generate word diff");
2150 free(minus.ptr);
2151 free(plus.ptr);
2152 if (diff_words->current_plus != diff_words->plus.text.ptr +
2153 diff_words->plus.text.size) {
2154 if (color_words_output_graph_prefix(diff_words))
2155 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2156 line_prefix, strlen(line_prefix), 0);
2157 fn_out_diff_words_write_helper(diff_words->opt,
2158 &style->ctx, style->newline,
2159 diff_words->plus.text.ptr + diff_words->plus.text.size
2160 - diff_words->current_plus, diff_words->current_plus);
2161 }
2162 diff_words->minus.text.size = diff_words->plus.text.size = 0;
2163 }
2164
2165 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2166 static void diff_words_flush(struct emit_callback *ecbdata)
2167 {
2168 struct diff_options *wo = ecbdata->diff_words->opt;
2169
2170 if (ecbdata->diff_words->minus.text.size ||
2171 ecbdata->diff_words->plus.text.size)
2172 diff_words_show(ecbdata->diff_words);
2173
2174 if (wo->emitted_symbols) {
2175 struct diff_options *o = ecbdata->opt;
2176 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2177 int i;
2178
2179 /*
2180 * NEEDSWORK:
2181 * Instead of appending each, concat all words to a line?
2182 */
2183 for (i = 0; i < wol->nr; i++)
2184 append_emitted_diff_symbol(o, &wol->buf[i]);
2185
2186 for (i = 0; i < wol->nr; i++)
2187 free((void *)wol->buf[i].line);
2188
2189 wol->nr = 0;
2190 }
2191 }
2192
2193 static void diff_filespec_load_driver(struct diff_filespec *one,
2194 struct index_state *istate)
2195 {
2196 /* Use already-loaded driver */
2197 if (one->driver)
2198 return;
2199
2200 if (S_ISREG(one->mode))
2201 one->driver = userdiff_find_by_path(istate, one->path);
2202
2203 /* Fallback to default settings */
2204 if (!one->driver)
2205 one->driver = userdiff_find_by_name("default");
2206 }
2207
2208 static const char *userdiff_word_regex(struct diff_filespec *one,
2209 struct index_state *istate)
2210 {
2211 diff_filespec_load_driver(one, istate);
2212 return one->driver->word_regex;
2213 }
2214
2215 static void init_diff_words_data(struct emit_callback *ecbdata,
2216 struct diff_options *orig_opts,
2217 struct diff_filespec *one,
2218 struct diff_filespec *two)
2219 {
2220 int i;
2221 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2222 memcpy(o, orig_opts, sizeof(struct diff_options));
2223
2224 ecbdata->diff_words =
2225 xcalloc(1, sizeof(struct diff_words_data));
2226 ecbdata->diff_words->type = o->word_diff;
2227 ecbdata->diff_words->opt = o;
2228
2229 if (orig_opts->emitted_symbols)
2230 o->emitted_symbols =
2231 xcalloc(1, sizeof(struct emitted_diff_symbols));
2232
2233 if (!o->word_regex)
2234 o->word_regex = userdiff_word_regex(one, o->repo->index);
2235 if (!o->word_regex)
2236 o->word_regex = userdiff_word_regex(two, o->repo->index);
2237 if (!o->word_regex)
2238 o->word_regex = diff_word_regex_cfg;
2239 if (o->word_regex) {
2240 ecbdata->diff_words->word_regex = (regex_t *)
2241 xmalloc(sizeof(regex_t));
2242 if (regcomp(ecbdata->diff_words->word_regex,
2243 o->word_regex,
2244 REG_EXTENDED | REG_NEWLINE))
2245 die("invalid regular expression: %s",
2246 o->word_regex);
2247 }
2248 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2249 if (o->word_diff == diff_words_styles[i].type) {
2250 ecbdata->diff_words->style =
2251 &diff_words_styles[i];
2252 break;
2253 }
2254 }
2255 if (want_color(o->use_color)) {
2256 struct diff_words_style *st = ecbdata->diff_words->style;
2257 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2258 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2259 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2260 }
2261 }
2262
2263 static void free_diff_words_data(struct emit_callback *ecbdata)
2264 {
2265 if (ecbdata->diff_words) {
2266 diff_words_flush(ecbdata);
2267 free (ecbdata->diff_words->opt->emitted_symbols);
2268 free (ecbdata->diff_words->opt);
2269 free (ecbdata->diff_words->minus.text.ptr);
2270 free (ecbdata->diff_words->minus.orig);
2271 free (ecbdata->diff_words->plus.text.ptr);
2272 free (ecbdata->diff_words->plus.orig);
2273 if (ecbdata->diff_words->word_regex) {
2274 regfree(ecbdata->diff_words->word_regex);
2275 free(ecbdata->diff_words->word_regex);
2276 }
2277 FREE_AND_NULL(ecbdata->diff_words);
2278 }
2279 }
2280
2281 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2282 {
2283 if (want_color(diff_use_color))
2284 return diff_colors[ix];
2285 return "";
2286 }
2287
2288 const char *diff_line_prefix(struct diff_options *opt)
2289 {
2290 struct strbuf *msgbuf;
2291 if (!opt->output_prefix)
2292 return "";
2293
2294 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2295 return msgbuf->buf;
2296 }
2297
2298 static unsigned long sane_truncate_line(char *line, unsigned long len)
2299 {
2300 const char *cp;
2301 unsigned long allot;
2302 size_t l = len;
2303
2304 cp = line;
2305 allot = l;
2306 while (0 < l) {
2307 (void) utf8_width(&cp, &l);
2308 if (!cp)
2309 break; /* truncated in the middle? */
2310 }
2311 return allot - l;
2312 }
2313
2314 static void find_lno(const char *line, struct emit_callback *ecbdata)
2315 {
2316 const char *p;
2317 ecbdata->lno_in_preimage = 0;
2318 ecbdata->lno_in_postimage = 0;
2319 p = strchr(line, '-');
2320 if (!p)
2321 return; /* cannot happen */
2322 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2323 p = strchr(p, '+');
2324 if (!p)
2325 return; /* cannot happen */
2326 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2327 }
2328
2329 static void fn_out_consume(void *priv, char *line, unsigned long len)
2330 {
2331 struct emit_callback *ecbdata = priv;
2332 struct diff_options *o = ecbdata->opt;
2333
2334 o->found_changes = 1;
2335
2336 if (ecbdata->header) {
2337 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2338 ecbdata->header->buf, ecbdata->header->len, 0);
2339 strbuf_reset(ecbdata->header);
2340 ecbdata->header = NULL;
2341 }
2342
2343 if (ecbdata->label_path[0]) {
2344 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2345 ecbdata->label_path[0],
2346 strlen(ecbdata->label_path[0]), 0);
2347 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2348 ecbdata->label_path[1],
2349 strlen(ecbdata->label_path[1]), 0);
2350 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2351 }
2352
2353 if (diff_suppress_blank_empty
2354 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2355 line[0] = '\n';
2356 len = 1;
2357 }
2358
2359 if (line[0] == '@') {
2360 if (ecbdata->diff_words)
2361 diff_words_flush(ecbdata);
2362 len = sane_truncate_line(line, len);
2363 find_lno(line, ecbdata);
2364 emit_hunk_header(ecbdata, line, len);
2365 return;
2366 }
2367
2368 if (ecbdata->diff_words) {
2369 enum diff_symbol s =
2370 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2371 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2372 if (line[0] == '-') {
2373 diff_words_append(line, len,
2374 &ecbdata->diff_words->minus);
2375 return;
2376 } else if (line[0] == '+') {
2377 diff_words_append(line, len,
2378 &ecbdata->diff_words->plus);
2379 return;
2380 } else if (starts_with(line, "\\ ")) {
2381 /*
2382 * Eat the "no newline at eof" marker as if we
2383 * saw a "+" or "-" line with nothing on it,
2384 * and return without diff_words_flush() to
2385 * defer processing. If this is the end of
2386 * preimage, more "+" lines may come after it.
2387 */
2388 return;
2389 }
2390 diff_words_flush(ecbdata);
2391 emit_diff_symbol(o, s, line, len, 0);
2392 return;
2393 }
2394
2395 switch (line[0]) {
2396 case '+':
2397 ecbdata->lno_in_postimage++;
2398 emit_add_line(ecbdata, line + 1, len - 1);
2399 break;
2400 case '-':
2401 ecbdata->lno_in_preimage++;
2402 emit_del_line(ecbdata, line + 1, len - 1);
2403 break;
2404 case ' ':
2405 ecbdata->lno_in_postimage++;
2406 ecbdata->lno_in_preimage++;
2407 emit_context_line(ecbdata, line + 1, len - 1);
2408 break;
2409 default:
2410 /* incomplete line at the end */
2411 ecbdata->lno_in_preimage++;
2412 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2413 line, len, 0);
2414 break;
2415 }
2416 }
2417
2418 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2419 {
2420 const char *old_name = a;
2421 const char *new_name = b;
2422 int pfx_length, sfx_length;
2423 int pfx_adjust_for_slash;
2424 int len_a = strlen(a);
2425 int len_b = strlen(b);
2426 int a_midlen, b_midlen;
2427 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2428 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2429
2430 if (qlen_a || qlen_b) {
2431 quote_c_style(a, name, NULL, 0);
2432 strbuf_addstr(name, " => ");
2433 quote_c_style(b, name, NULL, 0);
2434 return;
2435 }
2436
2437 /* Find common prefix */
2438 pfx_length = 0;
2439 while (*old_name && *new_name && *old_name == *new_name) {
2440 if (*old_name == '/')
2441 pfx_length = old_name - a + 1;
2442 old_name++;
2443 new_name++;
2444 }
2445
2446 /* Find common suffix */
2447 old_name = a + len_a;
2448 new_name = b + len_b;
2449 sfx_length = 0;
2450 /*
2451 * If there is a common prefix, it must end in a slash. In
2452 * that case we let this loop run 1 into the prefix to see the
2453 * same slash.
2454 *
2455 * If there is no common prefix, we cannot do this as it would
2456 * underrun the input strings.
2457 */
2458 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2459 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2460 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2461 *old_name == *new_name) {
2462 if (*old_name == '/')
2463 sfx_length = len_a - (old_name - a);
2464 old_name--;
2465 new_name--;
2466 }
2467
2468 /*
2469 * pfx{mid-a => mid-b}sfx
2470 * {pfx-a => pfx-b}sfx
2471 * pfx{sfx-a => sfx-b}
2472 * name-a => name-b
2473 */
2474 a_midlen = len_a - pfx_length - sfx_length;
2475 b_midlen = len_b - pfx_length - sfx_length;
2476 if (a_midlen < 0)
2477 a_midlen = 0;
2478 if (b_midlen < 0)
2479 b_midlen = 0;
2480
2481 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2482 if (pfx_length + sfx_length) {
2483 strbuf_add(name, a, pfx_length);
2484 strbuf_addch(name, '{');
2485 }
2486 strbuf_add(name, a + pfx_length, a_midlen);
2487 strbuf_addstr(name, " => ");
2488 strbuf_add(name, b + pfx_length, b_midlen);
2489 if (pfx_length + sfx_length) {
2490 strbuf_addch(name, '}');
2491 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2492 }
2493 }
2494
2495 struct diffstat_t {
2496 int nr;
2497 int alloc;
2498 struct diffstat_file {
2499 char *from_name;
2500 char *name;
2501 char *print_name;
2502 const char *comments;
2503 unsigned is_unmerged:1;
2504 unsigned is_binary:1;
2505 unsigned is_renamed:1;
2506 unsigned is_interesting:1;
2507 uintmax_t added, deleted;
2508 } **files;
2509 };
2510
2511 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2512 const char *name_a,
2513 const char *name_b)
2514 {
2515 struct diffstat_file *x;
2516 x = xcalloc(1, sizeof(*x));
2517 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2518 diffstat->files[diffstat->nr++] = x;
2519 if (name_b) {
2520 x->from_name = xstrdup(name_a);
2521 x->name = xstrdup(name_b);
2522 x->is_renamed = 1;
2523 }
2524 else {
2525 x->from_name = NULL;
2526 x->name = xstrdup(name_a);
2527 }
2528 return x;
2529 }
2530
2531 static void diffstat_consume(void *priv, char *line, unsigned long len)
2532 {
2533 struct diffstat_t *diffstat = priv;
2534 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2535
2536 if (line[0] == '+')
2537 x->added++;
2538 else if (line[0] == '-')
2539 x->deleted++;
2540 }
2541
2542 const char mime_boundary_leader[] = "------------";
2543
2544 static int scale_linear(int it, int width, int max_change)
2545 {
2546 if (!it)
2547 return 0;
2548 /*
2549 * make sure that at least one '-' or '+' is printed if
2550 * there is any change to this path. The easiest way is to
2551 * scale linearly as if the alloted width is one column shorter
2552 * than it is, and then add 1 to the result.
2553 */
2554 return 1 + (it * (width - 1) / max_change);
2555 }
2556
2557 static void show_graph(struct strbuf *out, char ch, int cnt,
2558 const char *set, const char *reset)
2559 {
2560 if (cnt <= 0)
2561 return;
2562 strbuf_addstr(out, set);
2563 strbuf_addchars(out, ch, cnt);
2564 strbuf_addstr(out, reset);
2565 }
2566
2567 static void fill_print_name(struct diffstat_file *file)
2568 {
2569 struct strbuf pname = STRBUF_INIT;
2570
2571 if (file->print_name)
2572 return;
2573
2574 if (file->is_renamed)
2575 pprint_rename(&pname, file->from_name, file->name);
2576 else
2577 quote_c_style(file->name, &pname, NULL, 0);
2578
2579 if (file->comments)
2580 strbuf_addf(&pname, " (%s)", file->comments);
2581
2582 file->print_name = strbuf_detach(&pname, NULL);
2583 }
2584
2585 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2586 int files, int insertions, int deletions)
2587 {
2588 struct strbuf sb = STRBUF_INIT;
2589
2590 if (!files) {
2591 assert(insertions == 0 && deletions == 0);
2592 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2593 NULL, 0, 0);
2594 return;
2595 }
2596
2597 strbuf_addf(&sb,
2598 (files == 1) ? " %d file changed" : " %d files changed",
2599 files);
2600
2601 /*
2602 * For binary diff, the caller may want to print "x files
2603 * changed" with insertions == 0 && deletions == 0.
2604 *
2605 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2606 * is probably less confusing (i.e skip over "2 files changed
2607 * but nothing about added/removed lines? Is this a bug in Git?").
2608 */
2609 if (insertions || deletions == 0) {
2610 strbuf_addf(&sb,
2611 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2612 insertions);
2613 }
2614
2615 if (deletions || insertions == 0) {
2616 strbuf_addf(&sb,
2617 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2618 deletions);
2619 }
2620 strbuf_addch(&sb, '\n');
2621 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2622 sb.buf, sb.len, 0);
2623 strbuf_release(&sb);
2624 }
2625
2626 void print_stat_summary(FILE *fp, int files,
2627 int insertions, int deletions)
2628 {
2629 struct diff_options o;
2630 memset(&o, 0, sizeof(o));
2631 o.file = fp;
2632
2633 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2634 }
2635
2636 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2637 {
2638 int i, len, add, del, adds = 0, dels = 0;
2639 uintmax_t max_change = 0, max_len = 0;
2640 int total_files = data->nr, count;
2641 int width, name_width, graph_width, number_width = 0, bin_width = 0;
2642 const char *reset, *add_c, *del_c;
2643 int extra_shown = 0;
2644 const char *line_prefix = diff_line_prefix(options);
2645 struct strbuf out = STRBUF_INIT;
2646
2647 if (data->nr == 0)
2648 return;
2649
2650 count = options->stat_count ? options->stat_count : data->nr;
2651
2652 reset = diff_get_color_opt(options, DIFF_RESET);
2653 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2654 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2655
2656 /*
2657 * Find the longest filename and max number of changes
2658 */
2659 for (i = 0; (i < count) && (i < data->nr); i++) {
2660 struct diffstat_file *file = data->files[i];
2661 uintmax_t change = file->added + file->deleted;
2662
2663 if (!file->is_interesting && (change == 0)) {
2664 count++; /* not shown == room for one more */
2665 continue;
2666 }
2667 fill_print_name(file);
2668 len = strlen(file->print_name);
2669 if (max_len < len)
2670 max_len = len;
2671
2672 if (file->is_unmerged) {
2673 /* "Unmerged" is 8 characters */
2674 bin_width = bin_width < 8 ? 8 : bin_width;
2675 continue;
2676 }
2677 if (file->is_binary) {
2678 /* "Bin XXX -> YYY bytes" */
2679 int w = 14 + decimal_width(file->added)
2680 + decimal_width(file->deleted);
2681 bin_width = bin_width < w ? w : bin_width;
2682 /* Display change counts aligned with "Bin" */
2683 number_width = 3;
2684 continue;
2685 }
2686
2687 if (max_change < change)
2688 max_change = change;
2689 }
2690 count = i; /* where we can stop scanning in data->files[] */
2691
2692 /*
2693 * We have width = stat_width or term_columns() columns total.
2694 * We want a maximum of min(max_len, stat_name_width) for the name part.
2695 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2696 * We also need 1 for " " and 4 + decimal_width(max_change)
2697 * for " | NNNN " and one the empty column at the end, altogether
2698 * 6 + decimal_width(max_change).
2699 *
2700 * If there's not enough space, we will use the smaller of
2701 * stat_name_width (if set) and 5/8*width for the filename,
2702 * and the rest for constant elements + graph part, but no more
2703 * than stat_graph_width for the graph part.
2704 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2705 * for the standard terminal size).
2706 *
2707 * In other words: stat_width limits the maximum width, and
2708 * stat_name_width fixes the maximum width of the filename,
2709 * and is also used to divide available columns if there
2710 * aren't enough.
2711 *
2712 * Binary files are displayed with "Bin XXX -> YYY bytes"
2713 * instead of the change count and graph. This part is treated
2714 * similarly to the graph part, except that it is not
2715 * "scaled". If total width is too small to accommodate the
2716 * guaranteed minimum width of the filename part and the
2717 * separators and this message, this message will "overflow"
2718 * making the line longer than the maximum width.
2719 */
2720
2721 if (options->stat_width == -1)
2722 width = term_columns() - strlen(line_prefix);
2723 else
2724 width = options->stat_width ? options->stat_width : 80;
2725 number_width = decimal_width(max_change) > number_width ?
2726 decimal_width(max_change) : number_width;
2727
2728 if (options->stat_graph_width == -1)
2729 options->stat_graph_width = diff_stat_graph_width;
2730
2731 /*
2732 * Guarantee 3/8*16==6 for the graph part
2733 * and 5/8*16==10 for the filename part
2734 */
2735 if (width < 16 + 6 + number_width)
2736 width = 16 + 6 + number_width;
2737
2738 /*
2739 * First assign sizes that are wanted, ignoring available width.
2740 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2741 * starting from "XXX" should fit in graph_width.
2742 */
2743 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2744 if (options->stat_graph_width &&
2745 options->stat_graph_width < graph_width)
2746 graph_width = options->stat_graph_width;
2747
2748 name_width = (options->stat_name_width > 0 &&
2749 options->stat_name_width < max_len) ?
2750 options->stat_name_width : max_len;
2751
2752 /*
2753 * Adjust adjustable widths not to exceed maximum width
2754 */
2755 if (name_width + number_width + 6 + graph_width > width) {
2756 if (graph_width > width * 3/8 - number_width - 6) {
2757 graph_width = width * 3/8 - number_width - 6;
2758 if (graph_width < 6)
2759 graph_width = 6;
2760 }
2761
2762 if (options->stat_graph_width &&
2763 graph_width > options->stat_graph_width)
2764 graph_width = options->stat_graph_width;
2765 if (name_width > width - number_width - 6 - graph_width)
2766 name_width = width - number_width - 6 - graph_width;
2767 else
2768 graph_width = width - number_width - 6 - name_width;
2769 }
2770
2771 /*
2772 * From here name_width is the width of the name area,
2773 * and graph_width is the width of the graph area.
2774 * max_change is used to scale graph properly.
2775 */
2776 for (i = 0; i < count; i++) {
2777 const char *prefix = "";
2778 struct diffstat_file *file = data->files[i];
2779 char *name = file->print_name;
2780 uintmax_t added = file->added;
2781 uintmax_t deleted = file->deleted;
2782 int name_len;
2783
2784 if (!file->is_interesting && (added + deleted == 0))
2785 continue;
2786
2787 /*
2788 * "scale" the filename
2789 */
2790 len = name_width;
2791 name_len = strlen(name);
2792 if (name_width < name_len) {
2793 char *slash;
2794 prefix = "...";
2795 len -= 3;
2796 name += name_len - len;
2797 slash = strchr(name, '/');
2798 if (slash)
2799 name = slash;
2800 }
2801
2802 if (file->is_binary) {
2803 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2804 strbuf_addf(&out, " %*s", number_width, "Bin");
2805 if (!added && !deleted) {
2806 strbuf_addch(&out, '\n');
2807 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2808 out.buf, out.len, 0);
2809 strbuf_reset(&out);
2810 continue;
2811 }
2812 strbuf_addf(&out, " %s%"PRIuMAX"%s",
2813 del_c, deleted, reset);
2814 strbuf_addstr(&out, " -> ");
2815 strbuf_addf(&out, "%s%"PRIuMAX"%s",
2816 add_c, added, reset);
2817 strbuf_addstr(&out, " bytes\n");
2818 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2819 out.buf, out.len, 0);
2820 strbuf_reset(&out);
2821 continue;
2822 }
2823 else if (file->is_unmerged) {
2824 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2825 strbuf_addstr(&out, " Unmerged\n");
2826 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2827 out.buf, out.len, 0);
2828 strbuf_reset(&out);
2829 continue;
2830 }
2831
2832 /*
2833 * scale the add/delete
2834 */
2835 add = added;
2836 del = deleted;
2837
2838 if (graph_width <= max_change) {
2839 int total = scale_linear(add + del, graph_width, max_change);
2840 if (total < 2 && add && del)
2841 /* width >= 2 due to the sanity check */
2842 total = 2;
2843 if (add < del) {
2844 add = scale_linear(add, graph_width, max_change);
2845 del = total - add;
2846 } else {
2847 del = scale_linear(del, graph_width, max_change);
2848 add = total - del;
2849 }
2850 }
2851 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2852 strbuf_addf(&out, " %*"PRIuMAX"%s",
2853 number_width, added + deleted,
2854 added + deleted ? " " : "");
2855 show_graph(&out, '+', add, add_c, reset);
2856 show_graph(&out, '-', del, del_c, reset);
2857 strbuf_addch(&out, '\n');
2858 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2859 out.buf, out.len, 0);
2860 strbuf_reset(&out);
2861 }
2862
2863 for (i = 0; i < data->nr; i++) {
2864 struct diffstat_file *file = data->files[i];
2865 uintmax_t added = file->added;
2866 uintmax_t deleted = file->deleted;
2867
2868 if (file->is_unmerged ||
2869 (!file->is_interesting && (added + deleted == 0))) {
2870 total_files--;
2871 continue;
2872 }
2873
2874 if (!file->is_binary) {
2875 adds += added;
2876 dels += deleted;
2877 }
2878 if (i < count)
2879 continue;
2880 if (!extra_shown)
2881 emit_diff_symbol(options,
2882 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2883 NULL, 0, 0);
2884 extra_shown = 1;
2885 }
2886
2887 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2888 strbuf_release(&out);
2889 }
2890
2891 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2892 {
2893 int i, adds = 0, dels = 0, total_files = data->nr;
2894
2895 if (data->nr == 0)
2896 return;
2897
2898 for (i = 0; i < data->nr; i++) {
2899 int added = data->files[i]->added;
2900 int deleted = data->files[i]->deleted;
2901
2902 if (data->files[i]->is_unmerged ||
2903 (!data->files[i]->is_interesting && (added + deleted == 0))) {
2904 total_files--;
2905 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2906 adds += added;
2907 dels += deleted;
2908 }
2909 }
2910 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2911 }
2912
2913 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2914 {
2915 int i;
2916
2917 if (data->nr == 0)
2918 return;
2919
2920 for (i = 0; i < data->nr; i++) {
2921 struct diffstat_file *file = data->files[i];
2922
2923 fprintf(options->file, "%s", diff_line_prefix(options));
2924
2925 if (file->is_binary)
2926 fprintf(options->file, "-\t-\t");
2927 else
2928 fprintf(options->file,
2929 "%"PRIuMAX"\t%"PRIuMAX"\t",
2930 file->added, file->deleted);
2931 if (options->line_termination) {
2932 fill_print_name(file);
2933 if (!file->is_renamed)
2934 write_name_quoted(file->name, options->file,
2935 options->line_termination);
2936 else {
2937 fputs(file->print_name, options->file);
2938 putc(options->line_termination, options->file);
2939 }
2940 } else {
2941 if (file->is_renamed) {
2942 putc('\0', options->file);
2943 write_name_quoted(file->from_name, options->file, '\0');
2944 }
2945 write_name_quoted(file->name, options->file, '\0');
2946 }
2947 }
2948 }
2949
2950 struct dirstat_file {
2951 const char *name;
2952 unsigned long changed;
2953 };
2954
2955 struct dirstat_dir {
2956 struct dirstat_file *files;
2957 int alloc, nr, permille, cumulative;
2958 };
2959
2960 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2961 unsigned long changed, const char *base, int baselen)
2962 {
2963 unsigned long sum_changes = 0;
2964 unsigned int sources = 0;
2965 const char *line_prefix = diff_line_prefix(opt);
2966
2967 while (dir->nr) {
2968 struct dirstat_file *f = dir->files;
2969 int namelen = strlen(f->name);
2970 unsigned long changes;
2971 char *slash;
2972
2973 if (namelen < baselen)
2974 break;
2975 if (memcmp(f->name, base, baselen))
2976 break;
2977 slash = strchr(f->name + baselen, '/');
2978 if (slash) {
2979 int newbaselen = slash + 1 - f->name;
2980 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2981 sources++;
2982 } else {
2983 changes = f->changed;
2984 dir->files++;
2985 dir->nr--;
2986 sources += 2;
2987 }
2988 sum_changes += changes;
2989 }
2990
2991 /*
2992 * We don't report dirstat's for
2993 * - the top level
2994 * - or cases where everything came from a single directory
2995 * under this directory (sources == 1).
2996 */
2997 if (baselen && sources != 1) {
2998 if (sum_changes) {
2999 int permille = sum_changes * 1000 / changed;
3000 if (permille >= dir->permille) {
3001 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3002 permille / 10, permille % 10, baselen, base);
3003 if (!dir->cumulative)
3004 return 0;
3005 }
3006 }
3007 }
3008 return sum_changes;
3009 }
3010
3011 static int dirstat_compare(const void *_a, const void *_b)
3012 {
3013 const struct dirstat_file *a = _a;
3014 const struct dirstat_file *b = _b;
3015 return strcmp(a->name, b->name);
3016 }
3017
3018 static void show_dirstat(struct diff_options *options)
3019 {
3020 int i;
3021 unsigned long changed;
3022 struct dirstat_dir dir;
3023 struct diff_queue_struct *q = &diff_queued_diff;
3024
3025 dir.files = NULL;
3026 dir.alloc = 0;
3027 dir.nr = 0;
3028 dir.permille = options->dirstat_permille;
3029 dir.cumulative = options->flags.dirstat_cumulative;
3030
3031 changed = 0;
3032 for (i = 0; i < q->nr; i++) {
3033 struct diff_filepair *p = q->queue[i];
3034 const char *name;
3035 unsigned long copied, added, damage;
3036
3037 name = p->two->path ? p->two->path : p->one->path;
3038
3039 if (p->one->oid_valid && p->two->oid_valid &&
3040 oideq(&p->one->oid, &p->two->oid)) {
3041 /*
3042 * The SHA1 has not changed, so pre-/post-content is
3043 * identical. We can therefore skip looking at the
3044 * file contents altogether.
3045 */
3046 damage = 0;
3047 goto found_damage;
3048 }
3049
3050 if (options->flags.dirstat_by_file) {
3051 /*
3052 * In --dirstat-by-file mode, we don't really need to
3053 * look at the actual file contents at all.
3054 * The fact that the SHA1 changed is enough for us to
3055 * add this file to the list of results
3056 * (with each file contributing equal damage).
3057 */
3058 damage = 1;
3059 goto found_damage;
3060 }
3061
3062 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3063 diff_populate_filespec(options->repo, p->one, 0);
3064 diff_populate_filespec(options->repo, p->two, 0);
3065 diffcore_count_changes(options->repo,
3066 p->one, p->two, NULL, NULL,
3067 &copied, &added);
3068 diff_free_filespec_data(p->one);
3069 diff_free_filespec_data(p->two);
3070 } else if (DIFF_FILE_VALID(p->one)) {
3071 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3072 copied = added = 0;
3073 diff_free_filespec_data(p->one);
3074 } else if (DIFF_FILE_VALID(p->two)) {
3075 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3076 copied = 0;
3077 added = p->two->size;
3078 diff_free_filespec_data(p->two);
3079 } else
3080 continue;
3081
3082 /*
3083 * Original minus copied is the removed material,
3084 * added is the new material. They are both damages
3085 * made to the preimage.
3086 * If the resulting damage is zero, we know that
3087 * diffcore_count_changes() considers the two entries to
3088 * be identical, but since the oid changed, we
3089 * know that there must have been _some_ kind of change,
3090 * so we force all entries to have damage > 0.
3091 */
3092 damage = (p->one->size - copied) + added;
3093 if (!damage)
3094 damage = 1;
3095
3096 found_damage:
3097 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3098 dir.files[dir.nr].name = name;
3099 dir.files[dir.nr].changed = damage;
3100 changed += damage;
3101 dir.nr++;
3102 }
3103
3104 /* This can happen even with many files, if everything was renames */
3105 if (!changed)
3106 return;
3107
3108 /* Show all directories with more than x% of the changes */
3109 QSORT(dir.files, dir.nr, dirstat_compare);
3110 gather_dirstat(options, &dir, changed, "", 0);
3111 }
3112
3113 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3114 {
3115 int i;
3116 unsigned long changed;
3117 struct dirstat_dir dir;
3118
3119 if (data->nr == 0)
3120 return;
3121
3122 dir.files = NULL;
3123 dir.alloc = 0;
3124 dir.nr = 0;
3125 dir.permille = options->dirstat_permille;
3126 dir.cumulative = options->flags.dirstat_cumulative;
3127
3128 changed = 0;
3129 for (i = 0; i < data->nr; i++) {
3130 struct diffstat_file *file = data->files[i];
3131 unsigned long damage = file->added + file->deleted;
3132 if (file->is_binary)
3133 /*
3134 * binary files counts bytes, not lines. Must find some
3135 * way to normalize binary bytes vs. textual lines.
3136 * The following heuristic assumes that there are 64
3137 * bytes per "line".
3138 * This is stupid and ugly, but very cheap...
3139 */
3140 damage = DIV_ROUND_UP(damage, 64);
3141 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3142 dir.files[dir.nr].name = file->name;
3143 dir.files[dir.nr].changed = damage;
3144 changed += damage;
3145 dir.nr++;
3146 }
3147
3148 /* This can happen even with many files, if everything was renames */
3149 if (!changed)
3150 return;
3151
3152 /* Show all directories with more than x% of the changes */
3153 QSORT(dir.files, dir.nr, dirstat_compare);
3154 gather_dirstat(options, &dir, changed, "", 0);
3155 }
3156
3157 static void free_diffstat_info(struct diffstat_t *diffstat)
3158 {
3159 int i;
3160 for (i = 0; i < diffstat->nr; i++) {
3161 struct diffstat_file *f = diffstat->files[i];
3162 free(f->print_name);
3163 free(f->name);
3164 free(f->from_name);
3165 free(f);
3166 }
3167 free(diffstat->files);
3168 }
3169
3170 struct checkdiff_t {
3171 const char *filename;
3172 int lineno;
3173 int conflict_marker_size;
3174 struct diff_options *o;
3175 unsigned ws_rule;
3176 unsigned status;
3177 };
3178
3179 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3180 {
3181 char firstchar;
3182 int cnt;
3183
3184 if (len < marker_size + 1)
3185 return 0;
3186 firstchar = line[0];
3187 switch (firstchar) {
3188 case '=': case '>': case '<': case '|':
3189 break;
3190 default:
3191 return 0;
3192 }
3193 for (cnt = 1; cnt < marker_size; cnt++)
3194 if (line[cnt] != firstchar)
3195 return 0;
3196 /* line[1] thru line[marker_size-1] are same as firstchar */
3197 if (len < marker_size + 1 || !isspace(line[marker_size]))
3198 return 0;
3199 return 1;
3200 }
3201
3202 static void checkdiff_consume_hunk(void *priv,
3203 long ob, long on, long nb, long nn,
3204 const char *func, long funclen)
3205
3206 {
3207 struct checkdiff_t *data = priv;
3208 data->lineno = nb - 1;
3209 }
3210
3211 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3212 {
3213 struct checkdiff_t *data = priv;
3214 int marker_size = data->conflict_marker_size;
3215 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3216 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3217 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3218 char *err;
3219 const char *line_prefix;
3220
3221 assert(data->o);
3222 line_prefix = diff_line_prefix(data->o);
3223
3224 if (line[0] == '+') {
3225 unsigned bad;
3226 data->lineno++;
3227 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3228 data->status |= 1;
3229 fprintf(data->o->file,
3230 "%s%s:%d: leftover conflict marker\n",
3231 line_prefix, data->filename, data->lineno);
3232 }
3233 bad = ws_check(line + 1, len - 1, data->ws_rule);
3234 if (!bad)
3235 return;
3236 data->status |= bad;
3237 err = whitespace_error_string(bad);
3238 fprintf(data->o->file, "%s%s:%d: %s.\n",
3239 line_prefix, data->filename, data->lineno, err);
3240 free(err);
3241 emit_line(data->o, set, reset, line, 1);
3242 ws_check_emit(line + 1, len - 1, data->ws_rule,
3243 data->o->file, set, reset, ws);
3244 } else if (line[0] == ' ') {
3245 data->lineno++;
3246 }
3247 }
3248
3249 static unsigned char *deflate_it(char *data,
3250 unsigned long size,
3251 unsigned long *result_size)
3252 {
3253 int bound;
3254 unsigned char *deflated;
3255 git_zstream stream;
3256
3257 git_deflate_init(&stream, zlib_compression_level);
3258 bound = git_deflate_bound(&stream, size);
3259 deflated = xmalloc(bound);
3260 stream.next_out = deflated;
3261 stream.avail_out = bound;
3262
3263 stream.next_in = (unsigned char *)data;
3264 stream.avail_in = size;
3265 while (git_deflate(&stream, Z_FINISH) == Z_OK)
3266 ; /* nothing */
3267 git_deflate_end(&stream);
3268 *result_size = stream.total_out;
3269 return deflated;
3270 }
3271
3272 static void emit_binary_diff_body(struct diff_options *o,
3273 mmfile_t *one, mmfile_t *two)
3274 {
3275 void *cp;
3276 void *delta;
3277 void *deflated;
3278 void *data;
3279 unsigned long orig_size;
3280 unsigned long delta_size;
3281 unsigned long deflate_size;
3282 unsigned long data_size;
3283
3284 /* We could do deflated delta, or we could do just deflated two,
3285 * whichever is smaller.
3286 */
3287 delta = NULL;
3288 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3289 if (one->size && two->size) {
3290 delta = diff_delta(one->ptr, one->size,
3291 two->ptr, two->size,
3292 &delta_size, deflate_size);
3293 if (delta) {
3294 void *to_free = delta;
3295 orig_size = delta_size;
3296 delta = deflate_it(delta, delta_size, &delta_size);
3297 free(to_free);
3298 }
3299 }
3300
3301 if (delta && delta_size < deflate_size) {
3302 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3303 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3304 s, strlen(s), 0);
3305 free(s);
3306 free(deflated);
3307 data = delta;
3308 data_size = delta_size;
3309 } else {
3310 char *s = xstrfmt("%lu", two->size);
3311 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3312 s, strlen(s), 0);
3313 free(s);
3314 free(delta);
3315 data = deflated;
3316 data_size = deflate_size;
3317 }
3318
3319 /* emit data encoded in base85 */
3320 cp = data;
3321 while (data_size) {
3322 int len;
3323 int bytes = (52 < data_size) ? 52 : data_size;
3324 char line[71];
3325 data_size -= bytes;
3326 if (bytes <= 26)
3327 line[0] = bytes + 'A' - 1;
3328 else
3329 line[0] = bytes - 26 + 'a' - 1;
3330 encode_85(line + 1, cp, bytes);
3331 cp = (char *) cp + bytes;
3332
3333 len = strlen(line);
3334 line[len++] = '\n';
3335 line[len] = '\0';
3336
3337 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3338 line, len, 0);
3339 }
3340 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3341 free(data);
3342 }
3343
3344 static void emit_binary_diff(struct diff_options *o,
3345 mmfile_t *one, mmfile_t *two)
3346 {
3347 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3348 emit_binary_diff_body(o, one, two);
3349 emit_binary_diff_body(o, two, one);
3350 }
3351
3352 int diff_filespec_is_binary(struct repository *r,
3353 struct diff_filespec *one)
3354 {
3355 if (one->is_binary == -1) {
3356 diff_filespec_load_driver(one, r->index);
3357 if (one->driver->binary != -1)
3358 one->is_binary = one->driver->binary;
3359 else {
3360 if (!one->data && DIFF_FILE_VALID(one))
3361 diff_populate_filespec(r, one, CHECK_BINARY);
3362 if (one->is_binary == -1 && one->data)
3363 one->is_binary = buffer_is_binary(one->data,
3364 one->size);
3365 if (one->is_binary == -1)
3366 one->is_binary = 0;
3367 }
3368 }
3369 return one->is_binary;
3370 }
3371
3372 static const struct userdiff_funcname *
3373 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3374 {
3375 diff_filespec_load_driver(one, o->repo->index);
3376 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3377 }
3378
3379 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3380 {
3381 if (!options->a_prefix)
3382 options->a_prefix = a;
3383 if (!options->b_prefix)
3384 options->b_prefix = b;
3385 }
3386
3387 struct userdiff_driver *get_textconv(struct repository *r,
3388 struct diff_filespec *one)
3389 {
3390 if (!DIFF_FILE_VALID(one))
3391 return NULL;
3392
3393 diff_filespec_load_driver(one, r->index);
3394 return userdiff_get_textconv(r, one->driver);
3395 }
3396
3397 static void builtin_diff(const char *name_a,
3398 const char *name_b,
3399 struct diff_filespec *one,
3400 struct diff_filespec *two,
3401 const char *xfrm_msg,
3402 int must_show_header,
3403 struct diff_options *o,
3404 int complete_rewrite)
3405 {
3406 mmfile_t mf1, mf2;
3407 const char *lbl[2];
3408 char *a_one, *b_two;
3409 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3410 const char *reset = diff_get_color_opt(o, DIFF_RESET);
3411 const char *a_prefix, *b_prefix;
3412 struct userdiff_driver *textconv_one = NULL;
3413 struct userdiff_driver *textconv_two = NULL;
3414 struct strbuf header = STRBUF_INIT;
3415 const char *line_prefix = diff_line_prefix(o);
3416
3417 diff_set_mnemonic_prefix(o, "a/", "b/");
3418 if (o->flags.reverse_diff) {
3419 a_prefix = o->b_prefix;
3420 b_prefix = o->a_prefix;
3421 } else {
3422 a_prefix = o->a_prefix;
3423 b_prefix = o->b_prefix;
3424 }
3425
3426 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3427 (!one->mode || S_ISGITLINK(one->mode)) &&
3428 (!two->mode || S_ISGITLINK(two->mode))) {
3429 show_submodule_summary(o, one->path ? one->path : two->path,
3430 &one->oid, &two->oid,
3431 two->dirty_submodule);
3432 return;
3433 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3434 (!one->mode || S_ISGITLINK(one->mode)) &&
3435 (!two->mode || S_ISGITLINK(two->mode))) {
3436 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3437 &one->oid, &two->oid,
3438 two->dirty_submodule);
3439 return;
3440 }
3441
3442 if (o->flags.allow_textconv) {
3443 textconv_one = get_textconv(o->repo, one);
3444 textconv_two = get_textconv(o->repo, two);
3445 }
3446
3447 /* Never use a non-valid filename anywhere if at all possible */
3448 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3449 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3450
3451 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3452 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3453 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3454 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3455 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3456 if (lbl[0][0] == '/') {
3457 /* /dev/null */
3458 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3459 if (xfrm_msg)
3460 strbuf_addstr(&header, xfrm_msg);
3461 must_show_header = 1;
3462 }
3463 else if (lbl[1][0] == '/') {
3464 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3465 if (xfrm_msg)
3466 strbuf_addstr(&header, xfrm_msg);
3467 must_show_header = 1;
3468 }
3469 else {
3470 if (one->mode != two->mode) {
3471 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3472 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3473 must_show_header = 1;
3474 }
3475 if (xfrm_msg)
3476 strbuf_addstr(&header, xfrm_msg);
3477
3478 /*
3479 * we do not run diff between different kind
3480 * of objects.
3481 */
3482 if ((one->mode ^ two->mode) & S_IFMT)
3483 goto free_ab_and_return;
3484 if (complete_rewrite &&
3485 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3486 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3487 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3488 header.buf, header.len, 0);
3489 strbuf_reset(&header);
3490 emit_rewrite_diff(name_a, name_b, one, two,
3491 textconv_one, textconv_two, o);
3492 o->found_changes = 1;
3493 goto free_ab_and_return;
3494 }
3495 }
3496
3497 if (o->irreversible_delete && lbl[1][0] == '/') {
3498 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3499 header.len, 0);
3500 strbuf_reset(&header);
3501 goto free_ab_and_return;
3502 } else if (!o->flags.text &&
3503 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3504 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3505 struct strbuf sb = STRBUF_INIT;
3506 if (!one->data && !two->data &&
3507 S_ISREG(one->mode) && S_ISREG(two->mode) &&
3508 !o->flags.binary) {
3509 if (oideq(&one->oid, &two->oid)) {
3510 if (must_show_header)
3511 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3512 header.buf, header.len,
3513 0);
3514 goto free_ab_and_return;
3515 }
3516 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3517 header.buf, header.len, 0);
3518 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3519 diff_line_prefix(o), lbl[0], lbl[1]);
3520 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3521 sb.buf, sb.len, 0);
3522 strbuf_release(&sb);
3523 goto free_ab_and_return;
3524 }
3525 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3526 fill_mmfile(o->repo, &mf2, two) < 0)
3527 die("unable to read files to diff");
3528 /* Quite common confusing case */
3529 if (mf1.size == mf2.size &&
3530 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3531 if (must_show_header)
3532 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3533 header.buf, header.len, 0);
3534 goto free_ab_and_return;
3535 }
3536 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3537 strbuf_reset(&header);
3538 if (o->flags.binary)
3539 emit_binary_diff(o, &mf1, &mf2);
3540 else {
3541 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3542 diff_line_prefix(o), lbl[0], lbl[1]);
3543 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3544 sb.buf, sb.len, 0);
3545 strbuf_release(&sb);
3546 }
3547 o->found_changes = 1;
3548 } else {
3549 /* Crazy xdl interfaces.. */
3550 const char *diffopts;
3551 const char *v;
3552 xpparam_t xpp;
3553 xdemitconf_t xecfg;
3554 struct emit_callback ecbdata;
3555 const struct userdiff_funcname *pe;
3556
3557 if (must_show_header) {
3558 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3559 header.buf, header.len, 0);
3560 strbuf_reset(&header);
3561 }
3562
3563 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3564 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3565
3566 pe = diff_funcname_pattern(o, one);
3567 if (!pe)
3568 pe = diff_funcname_pattern(o, two);
3569
3570 memset(&xpp, 0, sizeof(xpp));
3571 memset(&xecfg, 0, sizeof(xecfg));
3572 memset(&ecbdata, 0, sizeof(ecbdata));
3573 if (o->flags.suppress_diff_headers)
3574 lbl[0] = NULL;
3575 ecbdata.label_path = lbl;
3576 ecbdata.color_diff = want_color(o->use_color);
3577 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3578 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3579 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3580 ecbdata.opt = o;
3581 if (header.len && !o->flags.suppress_diff_headers)
3582 ecbdata.header = &header;
3583 xpp.flags = o->xdl_opts;
3584 xpp.anchors = o->anchors;
3585 xpp.anchors_nr = o->anchors_nr;
3586 xecfg.ctxlen = o->context;
3587 xecfg.interhunkctxlen = o->interhunkcontext;
3588 xecfg.flags = XDL_EMIT_FUNCNAMES;
3589 if (o->flags.funccontext)
3590 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3591 if (pe)
3592 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3593
3594 diffopts = getenv("GIT_DIFF_OPTS");
3595 if (!diffopts)
3596 ;
3597 else if (skip_prefix(diffopts, "--unified=", &v))
3598 xecfg.ctxlen = strtoul(v, NULL, 10);
3599 else if (skip_prefix(diffopts, "-u", &v))
3600 xecfg.ctxlen = strtoul(v, NULL, 10);
3601
3602 if (o->word_diff)
3603 init_diff_words_data(&ecbdata, o, one, two);
3604 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3605 &ecbdata, &xpp, &xecfg))
3606 die("unable to generate diff for %s", one->path);
3607 if (o->word_diff)
3608 free_diff_words_data(&ecbdata);
3609 if (textconv_one)
3610 free(mf1.ptr);
3611 if (textconv_two)
3612 free(mf2.ptr);
3613 xdiff_clear_find_func(&xecfg);
3614 }
3615
3616 free_ab_and_return:
3617 strbuf_release(&header);
3618 diff_free_filespec_data(one);
3619 diff_free_filespec_data(two);
3620 free(a_one);
3621 free(b_two);
3622 return;
3623 }
3624
3625 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3626 {
3627 if (!is_renamed) {
3628 if (p->status == DIFF_STATUS_ADDED) {
3629 if (S_ISLNK(p->two->mode))
3630 return "new +l";
3631 else if ((p->two->mode & 0777) == 0755)
3632 return "new +x";
3633 else
3634 return "new";
3635 } else if (p->status == DIFF_STATUS_DELETED)
3636 return "gone";
3637 }
3638 if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3639 return "mode -l";
3640 else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3641 return "mode +l";
3642 else if ((p->one->mode & 0777) == 0644 &&
3643 (p->two->mode & 0777) == 0755)
3644 return "mode +x";
3645 else if ((p->one->mode & 0777) == 0755 &&
3646 (p->two->mode & 0777) == 0644)
3647 return "mode -x";
3648 return NULL;
3649 }
3650
3651 static void builtin_diffstat(const char *name_a, const char *name_b,
3652 struct diff_filespec *one,
3653 struct diff_filespec *two,
3654 struct diffstat_t *diffstat,
3655 struct diff_options *o,
3656 struct diff_filepair *p)
3657 {
3658 mmfile_t mf1, mf2;
3659 struct diffstat_file *data;
3660 int same_contents;
3661 int complete_rewrite = 0;
3662
3663 if (!DIFF_PAIR_UNMERGED(p)) {
3664 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3665 complete_rewrite = 1;
3666 }
3667
3668 data = diffstat_add(diffstat, name_a, name_b);
3669 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3670 if (o->flags.stat_with_summary)
3671 data->comments = get_compact_summary(p, data->is_renamed);
3672
3673 if (!one || !two) {
3674 data->is_unmerged = 1;
3675 return;
3676 }
3677
3678 same_contents = oideq(&one->oid, &two->oid);
3679
3680 if (diff_filespec_is_binary(o->repo, one) ||
3681 diff_filespec_is_binary(o->repo, two)) {
3682 data->is_binary = 1;
3683 if (same_contents) {
3684 data->added = 0;
3685 data->deleted = 0;
3686 } else {
3687 data->added = diff_filespec_size(o->repo, two);
3688 data->deleted = diff_filespec_size(o->repo, one);
3689 }
3690 }
3691
3692 else if (complete_rewrite) {
3693 diff_populate_filespec(o->repo, one, 0);
3694 diff_populate_filespec(o->repo, two, 0);
3695 data->deleted = count_lines(one->data, one->size);
3696 data->added = count_lines(two->data, two->size);
3697 }
3698
3699 else if (!same_contents) {
3700 /* Crazy xdl interfaces.. */
3701 xpparam_t xpp;
3702 xdemitconf_t xecfg;
3703
3704 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3705 fill_mmfile(o->repo, &mf2, two) < 0)
3706 die("unable to read files to diff");
3707
3708 memset(&xpp, 0, sizeof(xpp));
3709 memset(&xecfg, 0, sizeof(xecfg));
3710 xpp.flags = o->xdl_opts;
3711 xpp.anchors = o->anchors;
3712 xpp.anchors_nr = o->anchors_nr;
3713 xecfg.ctxlen = o->context;
3714 xecfg.interhunkctxlen = o->interhunkcontext;
3715 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3716 diffstat_consume, diffstat, &xpp, &xecfg))
3717 die("unable to generate diffstat for %s", one->path);
3718 }
3719
3720 diff_free_filespec_data(one);
3721 diff_free_filespec_data(two);
3722 }